Clin Physiol Funct Imaging (2014)
doi: 10.1111/cpf.12132
Can a first-order exponential decay model fit heart rate recovery after resistance exercise? bio Y. Nakamura4, Rhenan Bartels-Ferreira1, Elder D. de Sousa2, Gabriela A. Trevizani1, Lilian P. Silva3, Fa 5 2 5 Claudia L. M. Forjaz , Jorge Roberto P. Lima and Tiago Pecßanha 1
Biomedical Engineering Program, COPPE, Federal University of Rio de Janeiro, Rio de Janeiro, 2Laboratory of Motor Assessment, Faculty of Physical Education and Sports, University of Juiz de Fora, 3Faculty of Physiotheraphy, Federal University of Juiz de Fora, Juiz de Fora, 4Department of Physical Education, State University of Londrina, Londrina, and 5Exercise Hemodynamic Laboratory, School of Physical Education and Sport, University of Sao Paulo, S~ao Paulo, Brazil
Summary Correspondence Tiago Pecßanha, Exercise Hemodynamic Laboratory, School of Physical Education and Sport, University of Sao Paulo, Av. Prof. Melo Moraes, 65 Butant~a - S~ao Paulo, SP - 05508-030, Brazil E-mail: [email protected]
Accepted for publication Received 23 October 2013; accepted 3 January 2014
Key words autonomic modulation; autonomic nervous system; exercise; heart rate; heart rate recovery; resistance training
The time-constant of postexercise heart rate recovery (HRRs) obtained by fitting heart rate decay curve by a first-order exponential fitting has being used to assess cardiac autonomic recovery after endurance exercise. The feasibility of this model was not tested after resistance exercise (RE). The aim of this study was to test the goodness of fit of the first-order exponential decay model to fit heart rate recovery (HRR) after RE. Ten healthy subjects participated in the study. The experimental sessions occurred in two separated days and consisted of performance of 1 set of 10 repetitions at 50% or 80% of the load achieved on the one-repetition maximum test [low-intensity (LI) and high-intensity (HI) sessions, respectively]. Heart rate (HR) was continuously registered before and during exercise and also for 10 min of recovery. A monoexponential equation was used to fit the HRR curve during the postexercise period using different time windows (i.e. 30, 60, 90, … 600 s). For each time window, (i) HRRs was calculated and (ii) variation of HR explained by the model (R2 goodness of fit index) was assessed. The HRRs showed stabilization from 360 and 420 s on LI and HI, respectively. Acceptable R2 values were observed from the 360 s on LI (R2 > 065) and at all tested time windows on HI (R2 > 075). In conclusion, this study showed that using a minimum length of monitoring (~420 s) HRR after RE can be adequately modelled by a first-order exponential fitting.
Introduction Postexercise heart rate recovery (HRR) reflects cardiac vagal reactivation and sympathetic withdrawal that occurs after the end of the exercise. Therefore, HRR allows assessing cardiac autonomic reorganization after a physiological disturbance, and it can be used to investigate autonomic changes associated with several diseases. There is a negative association between heart rate (HR) decrement after an exercise and overall mortality (Cole et al., 1999; Jouven et al., 2005). This information reveals the importance of HRR in the clinical setting. Aerobic exercise has been classically recommended for improving cardiovascular health (Garber et al., 2011). HRR after an aerobic exercise bout can be modelled by a first-order exponential model (Perini et al., 1989; Pierpont et al., 2000). A number of investigations have used the time-constant of HR decay during the postexercise period (HRRs), obtained by the first-order exponential fitting, to assess the effect of physical
training (Buchheit et al., 2008; Barak et al., 2011) and acute exercise challenges (Buchheit et al., 2009, 2010; Nakamura et al., 2009; Al Haddad et al., 2012) on HRR in health and pathological conditions. In recent years, resistance exercise (RE) has also become a popular mode of exercise training for health gains in both men and women (Garber et al., 2011). The main adaptation brought by the regular practice of this mode of exercise is an increase in muscle mass and strength (Kraemer et al., 2004; Holviala et al., 2012). Nevertheless, there is evidence regarding the positive effect of RE on postexercise cardiovascular function and regulation (Queiroz et al., 2010; Cucato et al., 2011). Few studies, however, addressed HR and cardiac autonomic responses immediately after RE, likely due to the lack of adequate methods for this purpose. Heffernan et al. (2006) analysed HR variability by spectral analysis after RE and found a significant reduction in cardiac vagal modulation during the postexercise period. Nevertheless, spectral analysis of HR
© 2014 Scandinavian Society of Clinical Physiology and Nuclear Medicine. Published by John Wiley & Sons Ltd
1
2 A first-order exponential decay model, R. B. Ferreira et al.
variability during exercise and/or recovery requires a complex signal processing and may present problems with low resolution of the spectra when analysing short segments of data or with signal stability, which may limit its use in the clinical setting. It is important therefore to propose simpler methods for the assessment of cardiac autonomic responses after RE. The evaluation of HRR by a monoexponential model may be a good choice, but its adequacy should be tested. Thus, the aim of this study was to test the goodness of fit of the first-order exponential decay curve after RE. Since Pierpont et al. (2000) showed an influence of the exercise intensity and of the time window for curve fitting on the HRR kinetics after aerobic exercise, this study compared results between two distinct exercise intensities and used different postexercise periods of curve decay analyses.
Methods Subjects Ten healthy male subjects (22 2 years, 243 14 kg m2) with previous experience in resistance training and without any orthopaedic limitations volunteered to the study. They were advised not to engage in physical exercises and not to drink caffeine- or alcohol-containing beverages during the 24 h preceding the tests. The subjects provided voluntary written informed consent to participate in this study, which was approved by the University Human Ethics Review Board and followed the recommendations from the Declaration of Helsinki.
RS800cxâ HR monitor (Polar Electroâ Ltd, Kempele, Finland; sampling frequency of 1000 Hz) was used for continuous recording of the resting (10 min), exercise (30 s) and recovery (10 min) RR intervals (RRi). During the whole protocol, subjects remained seated at the bench of the LP machine to prevent modifications on cardiac autonomic regulation due to alterations on body position (Buchheit et al., 2009). Heart rate recovery analysis Data acquired by the HR monitor were transmitted to the Polar Pro Trainerâ software (v. 5.0, Polar Inc., Kempele, Finland). Subsequently, the signals were imported by MATLABâ (The MathWorks Inc., Natick, MA, USA), and the reciprocal of RRi was firstly determined and then multiplied by 60 for getting instantaneous HR. The average of resting HR (HRrest) and HR at the end of exercise (HRpeak) was calculated. A monoexponencial equation was used to fit the HRR curve during the postexercise period (Fig. 1) using the ‘nlinfit’ function provided in MATLABâ software. The nonlinear least squares fitting was performed by the Levenberg–Marquardt
Preliminary tests Prior to the experiments, subjects visited the laboratory in three separate days. At the first day, they received the experimental information, signed the informed consent form, were asked about physical training and underwent the anthropometric assessment (height and body mass). At the second and third days, the volunteers were assessed for their maximal strength on the 45° leg press exercise (LP), through the one-repetition maximum (1RM) test. Prior to the test, 15–20 warm-up repetitions were performed with 50% of estimated maximal load. Then, the subjects performed five attempts of reaching the maximum load, with 3–5 min of recovery interspersing the trials. The highest load reached over the trials was considered for the subsequent exercise session prescription. Experimental session All subjects underwent two exercise sessions in two different days, in a random order. In each one, they performed 1 set of 10 repetitions in LP exercise. The workloads were set at 50% of 1RM (low intensity, LI) in one session and at 80% of 1RM (high intensity, HI) in the other. In each occasion, a metronome was used to control the execution rhythm at 3 s for each repetition. Throughout the experimental sessions, a Polar
Figure 1 (Upper Panel) Beat-to-beat values of heart rate (HR) during recovery after high-intensity(HI) exercise in an illustrative volunteer and the curve obtained by Eq. (1). (Lower Panel) Scatter diagram of the residuals fluctuations around zero following a non-systematic pattern.
© 2014 Scandinavian Society of Clinical Physiology and Nuclear Medicine. Published by John Wiley & Sons Ltd
A first-order exponential decay model, R. B. Ferreira et al. 3
algorithm based on Gauss–Newton method (Marquardt, 1963). This algorithm returns a vector of estimated coefficients by the nonlinear fit of the HR responses in function of time, using the model specified by the following equation: HRðtÞ ¼ HR0 þ HRD: eHRRs þ ei t
ð1Þ
Where HR0 is the asymptotic value of HR as t?∞ HRΔ is the difference between HRpeak and HR0 HRRs is the time-constant -ei is the residual Different time windows were used for nonlinear fit: 30, 60, 120, 180, 240, 300, 360, 420, 480, 540 and 600 s. On each time segment, HR decay constant was calculated and the variation of HR explained by the model was assessed by the R square goodness of fit index (R2). These calculations allowed finding out for how long RRi must be acquired to obtain consistent results. Statistical analysis Statistical analysis was performed using the computing environment R (The R Foundation for Statistical Computing, Vienna, Austria). The results are reported as mean standard deviation. As the Lilliefors test showed that the data did not have normal distribution (Lilliefors, 1967), the Wilcoxon test was performed for comparing HRrest, HRpeak, HRRs and R² between the two sessions. Significance was set at P
doi: 10.1111/cpf.12132
Can a first-order exponential decay model fit heart rate recovery after resistance exercise? bio Y. Nakamura4, Rhenan Bartels-Ferreira1, Elder D. de Sousa2, Gabriela A. Trevizani1, Lilian P. Silva3, Fa 5 2 5 Claudia L. M. Forjaz , Jorge Roberto P. Lima and Tiago Pecßanha 1
Biomedical Engineering Program, COPPE, Federal University of Rio de Janeiro, Rio de Janeiro, 2Laboratory of Motor Assessment, Faculty of Physical Education and Sports, University of Juiz de Fora, 3Faculty of Physiotheraphy, Federal University of Juiz de Fora, Juiz de Fora, 4Department of Physical Education, State University of Londrina, Londrina, and 5Exercise Hemodynamic Laboratory, School of Physical Education and Sport, University of Sao Paulo, S~ao Paulo, Brazil
Summary Correspondence Tiago Pecßanha, Exercise Hemodynamic Laboratory, School of Physical Education and Sport, University of Sao Paulo, Av. Prof. Melo Moraes, 65 Butant~a - S~ao Paulo, SP - 05508-030, Brazil E-mail: [email protected]
Accepted for publication Received 23 October 2013; accepted 3 January 2014
Key words autonomic modulation; autonomic nervous system; exercise; heart rate; heart rate recovery; resistance training
The time-constant of postexercise heart rate recovery (HRRs) obtained by fitting heart rate decay curve by a first-order exponential fitting has being used to assess cardiac autonomic recovery after endurance exercise. The feasibility of this model was not tested after resistance exercise (RE). The aim of this study was to test the goodness of fit of the first-order exponential decay model to fit heart rate recovery (HRR) after RE. Ten healthy subjects participated in the study. The experimental sessions occurred in two separated days and consisted of performance of 1 set of 10 repetitions at 50% or 80% of the load achieved on the one-repetition maximum test [low-intensity (LI) and high-intensity (HI) sessions, respectively]. Heart rate (HR) was continuously registered before and during exercise and also for 10 min of recovery. A monoexponential equation was used to fit the HRR curve during the postexercise period using different time windows (i.e. 30, 60, 90, … 600 s). For each time window, (i) HRRs was calculated and (ii) variation of HR explained by the model (R2 goodness of fit index) was assessed. The HRRs showed stabilization from 360 and 420 s on LI and HI, respectively. Acceptable R2 values were observed from the 360 s on LI (R2 > 065) and at all tested time windows on HI (R2 > 075). In conclusion, this study showed that using a minimum length of monitoring (~420 s) HRR after RE can be adequately modelled by a first-order exponential fitting.
Introduction Postexercise heart rate recovery (HRR) reflects cardiac vagal reactivation and sympathetic withdrawal that occurs after the end of the exercise. Therefore, HRR allows assessing cardiac autonomic reorganization after a physiological disturbance, and it can be used to investigate autonomic changes associated with several diseases. There is a negative association between heart rate (HR) decrement after an exercise and overall mortality (Cole et al., 1999; Jouven et al., 2005). This information reveals the importance of HRR in the clinical setting. Aerobic exercise has been classically recommended for improving cardiovascular health (Garber et al., 2011). HRR after an aerobic exercise bout can be modelled by a first-order exponential model (Perini et al., 1989; Pierpont et al., 2000). A number of investigations have used the time-constant of HR decay during the postexercise period (HRRs), obtained by the first-order exponential fitting, to assess the effect of physical
training (Buchheit et al., 2008; Barak et al., 2011) and acute exercise challenges (Buchheit et al., 2009, 2010; Nakamura et al., 2009; Al Haddad et al., 2012) on HRR in health and pathological conditions. In recent years, resistance exercise (RE) has also become a popular mode of exercise training for health gains in both men and women (Garber et al., 2011). The main adaptation brought by the regular practice of this mode of exercise is an increase in muscle mass and strength (Kraemer et al., 2004; Holviala et al., 2012). Nevertheless, there is evidence regarding the positive effect of RE on postexercise cardiovascular function and regulation (Queiroz et al., 2010; Cucato et al., 2011). Few studies, however, addressed HR and cardiac autonomic responses immediately after RE, likely due to the lack of adequate methods for this purpose. Heffernan et al. (2006) analysed HR variability by spectral analysis after RE and found a significant reduction in cardiac vagal modulation during the postexercise period. Nevertheless, spectral analysis of HR
© 2014 Scandinavian Society of Clinical Physiology and Nuclear Medicine. Published by John Wiley & Sons Ltd
1
2 A first-order exponential decay model, R. B. Ferreira et al.
variability during exercise and/or recovery requires a complex signal processing and may present problems with low resolution of the spectra when analysing short segments of data or with signal stability, which may limit its use in the clinical setting. It is important therefore to propose simpler methods for the assessment of cardiac autonomic responses after RE. The evaluation of HRR by a monoexponential model may be a good choice, but its adequacy should be tested. Thus, the aim of this study was to test the goodness of fit of the first-order exponential decay curve after RE. Since Pierpont et al. (2000) showed an influence of the exercise intensity and of the time window for curve fitting on the HRR kinetics after aerobic exercise, this study compared results between two distinct exercise intensities and used different postexercise periods of curve decay analyses.
Methods Subjects Ten healthy male subjects (22 2 years, 243 14 kg m2) with previous experience in resistance training and without any orthopaedic limitations volunteered to the study. They were advised not to engage in physical exercises and not to drink caffeine- or alcohol-containing beverages during the 24 h preceding the tests. The subjects provided voluntary written informed consent to participate in this study, which was approved by the University Human Ethics Review Board and followed the recommendations from the Declaration of Helsinki.
RS800cxâ HR monitor (Polar Electroâ Ltd, Kempele, Finland; sampling frequency of 1000 Hz) was used for continuous recording of the resting (10 min), exercise (30 s) and recovery (10 min) RR intervals (RRi). During the whole protocol, subjects remained seated at the bench of the LP machine to prevent modifications on cardiac autonomic regulation due to alterations on body position (Buchheit et al., 2009). Heart rate recovery analysis Data acquired by the HR monitor were transmitted to the Polar Pro Trainerâ software (v. 5.0, Polar Inc., Kempele, Finland). Subsequently, the signals were imported by MATLABâ (The MathWorks Inc., Natick, MA, USA), and the reciprocal of RRi was firstly determined and then multiplied by 60 for getting instantaneous HR. The average of resting HR (HRrest) and HR at the end of exercise (HRpeak) was calculated. A monoexponencial equation was used to fit the HRR curve during the postexercise period (Fig. 1) using the ‘nlinfit’ function provided in MATLABâ software. The nonlinear least squares fitting was performed by the Levenberg–Marquardt
Preliminary tests Prior to the experiments, subjects visited the laboratory in three separate days. At the first day, they received the experimental information, signed the informed consent form, were asked about physical training and underwent the anthropometric assessment (height and body mass). At the second and third days, the volunteers were assessed for their maximal strength on the 45° leg press exercise (LP), through the one-repetition maximum (1RM) test. Prior to the test, 15–20 warm-up repetitions were performed with 50% of estimated maximal load. Then, the subjects performed five attempts of reaching the maximum load, with 3–5 min of recovery interspersing the trials. The highest load reached over the trials was considered for the subsequent exercise session prescription. Experimental session All subjects underwent two exercise sessions in two different days, in a random order. In each one, they performed 1 set of 10 repetitions in LP exercise. The workloads were set at 50% of 1RM (low intensity, LI) in one session and at 80% of 1RM (high intensity, HI) in the other. In each occasion, a metronome was used to control the execution rhythm at 3 s for each repetition. Throughout the experimental sessions, a Polar
Figure 1 (Upper Panel) Beat-to-beat values of heart rate (HR) during recovery after high-intensity(HI) exercise in an illustrative volunteer and the curve obtained by Eq. (1). (Lower Panel) Scatter diagram of the residuals fluctuations around zero following a non-systematic pattern.
© 2014 Scandinavian Society of Clinical Physiology and Nuclear Medicine. Published by John Wiley & Sons Ltd
A first-order exponential decay model, R. B. Ferreira et al. 3
algorithm based on Gauss–Newton method (Marquardt, 1963). This algorithm returns a vector of estimated coefficients by the nonlinear fit of the HR responses in function of time, using the model specified by the following equation: HRðtÞ ¼ HR0 þ HRD: eHRRs þ ei t
ð1Þ
Where HR0 is the asymptotic value of HR as t?∞ HRΔ is the difference between HRpeak and HR0 HRRs is the time-constant -ei is the residual Different time windows were used for nonlinear fit: 30, 60, 120, 180, 240, 300, 360, 420, 480, 540 and 600 s. On each time segment, HR decay constant was calculated and the variation of HR explained by the model was assessed by the R square goodness of fit index (R2). These calculations allowed finding out for how long RRi must be acquired to obtain consistent results. Statistical analysis Statistical analysis was performed using the computing environment R (The R Foundation for Statistical Computing, Vienna, Austria). The results are reported as mean standard deviation. As the Lilliefors test showed that the data did not have normal distribution (Lilliefors, 1967), the Wilcoxon test was performed for comparing HRrest, HRpeak, HRRs and R² between the two sessions. Significance was set at P
