Translational Stroke Research

, Volume 4, Issue 5, pp 488–499

Heart Rate Variability in Stroke Patients Submitted to an Acute Bout of Aerobic Exercise


    • Faculdade Medicina ABC
  • Luiz Carlos de Abreu
    • Faculdade Medicina ABC
  • Fernando Adami
    • Faculdade Medicina ABC
  • Franciele Marques Vanderlei
    • Faculdade Medicina ABC
  • Tatiana Dias de Carvalho
    • Faculdade Medicina ABC
  • Isadora Lessa Moreno
    • Faculdade Medicina ABC
  • Valdelias Xavier Pereira
    • Faculdade Medicina ABC
  • Vitor Engracia Valenti
    • Faculdade Medicina ABC
  • Monica Akemi Sato
    • Faculdade Medicina ABC
Original Article

DOI: 10.1007/s12975-013-0263-4

Cite this article as:
Raimundo, R.D., de Abreu, L.C., Adami, F. et al. Transl. Stroke Res. (2013) 4: 488. doi:10.1007/s12975-013-0263-4


Stroke has been associated with cardiac autonomic impairment due to damage in central nervous system. Dysfunction in heart rate variability (HRV) may reflect dysfunction of the autonomic nervous system. Aerobic training has been used in the rehabilitation procedure of patients, due to improvement of aerobic function and other beneficial effects as increased recruitment of motor units, favoring the development of muscle fibers. The purpose of this study was to evaluate the cardiac autonomic modulation in patients with stroke before, during, and after an acute bout of aerobic exercise. The heart rate of 38 stroke patients was recorded using a heart rate (HR) monitor and the data were used to assess cardiac autonomic modulation through HRV analysis. The patients were in supine position and remained at resting condition (R) for 10 min before starting the experiment. Afterwards, they were submitted to walking exercise (E) on a treadmill until achieve 50–70 % of maximum heart rate. After 30 min of aerobic exercise, the subjects were advised to remain in supine position for additional 30 min in order to record the HR during the recovery (RC) period. The recordings were divided in three periods: RC1, immediately after the end of exercise bout, RC2, between 12 and 17 min of recovery, and RC3, at the final 5 min of recovery. A significant decrease was observed during exercise in the MeanRR index (577.3 ± 92 vs. 861.1 + 109), RRtri (5.1 ± 2 vs. 9.1 ± 3), high frequency component (11.2 ± 4 vs. 167 ± 135 ms) and SD1 (5.7 ± 2 vs. 16.9 ± 7 ms) compared to resting values. The SDNN index reduced during E (27.6 ± 19) and RC1 (29.9 ± 11), RC2 (27.9 ± 9) and RC3 (32.4 ± 13) compared to resting values (42.4 ± 19). The low frequency component increased during E (545 ± 82), but decreased during RC1 (166.3 ± 129), RC2 (206.9 ± 152), and RC3 (249.5 ± 236) compared to R levels (394.6 ± 315). These findings suggest that stroke patients showed a reduced HRV during and at least 30 min after exercise, due to an autonomic imbalance reflected by increased indexes that represent the sympathetic nervous system.


Cardiac autonomic modulationHeart rateNeurological disordersCardiovascular responsesIschemiaVagus nerveAutonomic nervous systemThe sympathetic nervous systemThe parasympathetic nervous system


Stroke has been associated with impaired cardiac regulation, which may be linked to autonomic nervous system impairment [1]. The main clinical problems are linked to abnormalities in heart rate and changes in blood pressure (BP) due to an increase in sympathetic activity [1, 2]. The analysis of heart rate variability (HRV) is an investigative and noninvasive method well recognized in assessing autonomic modulation of the heart. Measurements performed in the time domain and frequency domain derived relevant information regarding the overall HRV as well as sympathetic and parasympathetic control of the heart [3, 4].

Heart rate variability abnormalities may reflect dysfunction of the autonomic nervous system (ANS). Changes in patterns of HRV provide a sensitive and early indicator of health impairments. High HRV is a sign of good adaptation while low HRV is often an indicator of abnormal and insufficient adaptation of ANS, which may indicate the presence of physiological malfunction in the individual, requiring further investigations in order to find a specific diagnosis. The reduction in HRV in resting conditions may be used as a predictor of cardiovascular risk, disability in cardiac baroreflex sensitivity with a worse prognosis in patients after stroke [46].

Cerebrovascular diseases may change autonomic function and increase cardiac events due to increased plasma catecholamines [5, 6]. These changes are linked to an increased susceptibility to sudden death. In stroke patients, the arrhythmic events are responsible for 6 % of sudden death. Previous studies performed in animals suggested that the right hemisphere strokes produce greater sympathetic activation than the left side [68].

Approximately 50 to 70 % of stroke survivors acquire functional independence, nevertheless, 30 to 50 % become permanently disabled, and among those few patients need to be assisted with personal care. One of the main changes observed in these individuals is motor weakness that compromises the functional capacity and mainly gait [9]. The weakness of the muscles that promote knee extension may be used as a predictor of risk for impaired walking ability [10]. Several studies [915] demonstrated the importance of strength exercises to improve gait. However, most studies use physical training programs of low intensity due to the delicate clinical situation in post stroke patients [14, 15].

Aerobic training is used in the rehabilitation procedures of these patients due to low aerobic capacity caused by the decreased functional activity. Aerobic activity has the potential to increase the recruitment of motor units favoring the development of muscle fibers, besides providing a decrease in body weight, blood pressure reduction and decrease in total cholesterol and increase in HDL-cholesterol [1113].

It is necessary to understand the evolution of these patients during aerobic exercise seeking for noninvasive methods that may provide an improved care of them in order to make a safe choice of exercise prescription in this group of patients. This would justify the requirement of monitoring the HRV in those patients. Therefore, we aimed to evaluate the heart rate variability in stroke patients before, during and after an acute bout of aerobic exercise.

Materials and Methods

Study Population

The study followed a cross-sectional design that assessed the period between May 2012 and April 2013 in 38 patients (58.4 ± 8 years—65.8 % male), with the diagnosis of stroke, evidenced by imaging examination with medical report and time of injury above a year and considered insufficiently active by the International Physical Activity Questionnaire [14].

The study was approved by the Ethics Committee in Research of the School of Medicine of ABC (Protocol No. 33353) and all volunteers signed an informed consent form.

We did not included patients who had no conditions or gear motor to perform the exercise on a treadmill, which did not achieve the maximum heart rate (HR) during the stipulated time of aerobic exercise. Patients with other neurological diseases, acute heart failure, diabetes mellitus, chronic atrial fibrillation, sick sinus syndrome, atrioventricular block grade II or III, smokers, drinkers who used beta-blockers or antiarrhythmic medication and those who did not present medical authorization and prescription for physical activity.

Initial Assessment

Data were collected under stable conditions of temperature (21–23 °C) and humidity (40–60 %), and the environment was kept quiet and controlled. All equipments used in data collection were previously tested and adjusted by minimizing the chance of occurrence of practical problems. Moreover, the experimental procedures were performed between 1 and 4 P.M. to standardize the protocol [3, 4].

Initially, the patients were weighed without shoes and with minimal clothing as possible, in electronic scale (Toledo 2096PP, Sao Paulo, Brazil), with capacity up to 150 kg and subdivisions up to 50 g. Height was measured according to the Frankfurt plane. We used a wall stadiometer (WCS 2.20 m Wood, Sao Paulo, Brazil) attached to the wall, with subdivisions in centimeters and millimeters.

To better characterize the population of the study, patients underwent three physical and cognitive tests: Mini-Mental State Examination (MMSE) [15], Fugl-Meyer [16, 17], and Orpington [18, 19]. The MMSE aimed to assess specific cognitive functions: orientation as to time, advising the local word registration, attention and calculation, recall of words, language and visual constructive ability. The MMSE score range from a minimum of zero to a maximum of 30 points [15]. The Fugl-Meyer test was also used to assess physical performance through aspects of motor control (range of motion, pain, motor impairment of upper and lower limbs and balance), and data were scored from zero to three points for each item and it was added to the final [16, 17]. Finally, the Orpington prognostic scale was used to measure disability due to cerebral vascular event from the evaluation of upper limb motor deficits, proprioception, balance and cognition. The scores range from 1.6 to 6.8, characterizing the vascular event as mild, moderate and severe [18, 19].

The systolic blood (SBP) and diastolic pressure (DBP) were obtained by a single indirect measurement performed by a sphygmomanometer (Welch Allyn Tycos DS44 DuraShock, New York, USA), positioned on the left arm of the patient. Auscultation of the arterial pulse was performed using a stethoscope (Classic II 3M Littmann-, New York, USA) positioned on the middle region of the antecubital fossa, using a criteria established by the VI Brazilian Hypertension Guidelines [20]. HR was recorded beat-to-beat based on means of a heart rate monitor (S810i, Polar Electro Oy, Kempele, Finland) that was attached to the patient’s wrist and received signals from the band pickup, fixed on the precordial region. In addition, respiratory rate (RR) was measured by counting the breaths taken per minute, and oxygen saturation (SAT) measured using a finger oximeter portable (9500, Onix-Indumeda, Nonin Fernbrook Lane North Plymouth, USA) positioned on the third finger of the patient. All cardiopulmonary measurements were performed by a single evaluator with extensive experience in this type of procedure.

Experimental Protocol

Individuals were initially in the supine position and remained at rest for 10 min. After that, we measured resting heart rate The subjects were advised to walk on a treadmill (RT250pro, Movement, Sao Paulo, Brazil) for aerobic exercise divided in 5 min of warm up and 25 min of exercise, achieving 50–70 % of the maximum heart rate [2125].

The cardiorespiratory parameters SBP, DBP, HR, RR (respiratory rate), and SAT (partial oxygen saturation) were measured at resting, at the beginning and end of exercise.

After 30 min of aerobic exercise, the subjects were instructed to remain in supine position for additional 30 min. During the recovery period after exercise, the cardiorespiratory parameters were measured again.

Analysis of Heart Rate Variability

Heart rate was recorded beat to beat through the heart rate monitor with sampling frequency of 1,000 Hz (S810i, Polar Electro Oy, Kempele, Finland) and from these records, time series were generated and used for HRV analysis. During the period of greatest signal stability (during which time series beat-to-beat with HR values appeared stable and free from large transients), a range with a series of over 256 RR intervals was selected from RR intervals selected, these have been interpolated based on the method of Fast Fourier transformation (Welch periodogram: 256 points with overlapping 50 % and Hanning window), two bands being chosen for spectral analysis in the frequency domain (low frequency (LF) 0.04 to 0.15, high frequency (HF) 0.15 to 0.4 Hz) [2628].

The time series of RR intervals was subjected to digital filtering software by Polar Precision Performance SW (version 4.01.029) supplemented by manual to eliminate premature ectopic beats and artifacts, and only series with more than 95 % of sinus beats were included in the study [26]. The Kubios software version 2.0 was used for analysis of HRV indexes.

In the time domain, we used for statistical analysis of HRV the MeanRR, RMSSD, NN50, pNN50 and SDNN indexes.

The MeanRR index represents the mean of all values of analyzed cardiac interval series.

The RMSSD index is defined as the square root of the mean squared differences between adjacent normal RR intervals in an interval of time reflecting predominantly vagal cardiac modulation [29].

The pNN50 index represents the cardiac parasympathetic modulation, defined as the percentage of successive RR interval differences whose absolute value exceeds 50 ms [4, 30, 31].

The SDNN index, which reflects the total variability of RR intervals with participation of both branches of the ANS, represents the standard deviation of all normal RR intervals, expressed in milliseconds [4, 30, 31].

The analysis of HRV can also be done by using the following geometric methods: triangular index (RRtri), triangular interpolation of RR intervals (TINN) and Poincaré plot, which allow the present RR intervals in geometric patterns and approaches to assess the use HRV [3234].

The triangular index is calculated from the construction of a density histogram of RR intervals, which shows on the horizontal axis (x axis), the length of RR intervals, and the vertical axis (y axis), how often each occurred. The union of the points of the histogram columns forms a shape like a triangle and the width of the base of the triangle expresses the variability of RR intervals. The triangular index may be calculated by dividing the area (corresponding to the total number of RR intervals) by height (corresponds to the number of RR intervals with modal frequency) of the triangle [3234].

The TINN constitutes the baseline width distribution measured as the base of a triangle approximating the distribution of all RR intervals, and the difference of least squares was used to determine the triangle [3234].

The Poincaré graph allows each RR interval plotted against the previous interval, and for quantitative analysis of the graph, the following indexes were calculated: SD1 (standard deviation of instantaneous beat-to-beat variability), SD2 (standard deviation of long-term continuous RR intervals) and the SD1/SD2 ratio [35].

The variability in the frequency domain calculation was made by beat-to-beat and interpolated their spectra calculated by FFT. These spectra were then embedded in LF and HF which is the power of the LF band related essentially to sympathetic modulation on the heart, whereas the potency of the HF band related to vagal modulation of the heart, and cardiac sympathetic-vagal balance may be represented by the LF/HF ratio.

For analysis of the variability of cardiac interval in the frequency domain, we used the spectral components of LF and HF in milliseconds squared (ms2) and the ratio between these components (LF/HF) which represents the relative value of each spectral component in relation to the total power minus the components of very low frequency (VLF). For this analysis, the spectra were integrated into bands of LF and HF; LF from 0.04 to 0.15 Hz and HF ranging from 0.15 to 0.4 Hz. Spectral analysis was calculated using the algorithm of fast Fourier transformation [15].

Normalizing the data of spectral analysis may be used to minimize the effects of changes in the VLF band, but due to the short records this work was impossible to make a proper quantification of VLF and, therefore, LF/HF ration was considered related to cardiac sympathetic–parasympathetic balance. This is determined by dividing the potency of a particular component (LF or HF) spectrum of the total power minus the VLF component and multiplied by 100 [3, 4, 3235].

Heart rate variability was studied at the following times: “Resting” (R) which corresponded to 10 min of rest in the supine who served for analysis of heart rate variability (HRV) (baseline); “Exercise” (E) comprising the period exercise that had stabilization of heart rate training, only been sent for analysis the period of aerobic exercise by removing the first 5 min of exercise and the final minutes that heart rate began to decline due to deceleration of the treadmill. The third period of data collection which included the 30-min supine postexercise (Recovery Period) that was subdivided into three phases, “Recovery Period 1” (RC1) which corresponded to the first 5 min of recovery; “Recovery Period 2” (RC2) which corresponded to half of the recovery time, the period between the twelfth- (12th) and seventeenth- (17th) minute recovery; “Recovery Period 3” (RC3) which corresponded to the final 5 min of recovery, that corresponds to the twenty-fifth (25th) to thirtieth (30th) minutes (Fig. 1).
Fig. 1

Schematic representation of the moments of the study with the total time of collection

Statistical Analysis

Data are as mean ± SD. Excel programs and SPSS (Statistical Package for Social Research) version 17.0 for statistical analysis were used to prepare the database. Descriptive statistics were performed by measures of central tendency and dispersion. The Shapiro–Wilk test was used to verify data normality. To compare the index between the moments we applied the Friedmann test or one-way ANOVA with repeated measures according to the normality of the data. The significance level was set at p < 0.05. All values of p refer to comparison with resting (R) condition.


The characteristics of the subject as well as the cardiorespiratory parameters of the study population are presented in Tables 1, 2, and 3.
Table 1

Characteristics of the study population according to sex, type, and side of injury



N (%)






Type of injury





Side of Injury





N number, % Percentage

Table 2

Characteristics of the study population in terms of age, weight, time of injury, and neurological scales


Mean + SD

Age (years)

58.5 + 8

Weight (kilogram)

68.6 ± 6

Time of injury

5.3 ± 1


21.8 ± 5


26.1 ± 5


19.7 ± 3

FM balance

8.9 ± 1

FM sensibility

19.7 ± 3


35.3 ± 4

FM pain

42.0 ± 1

FM total

151.6 ± 10


2.7 ± 0.6

SD Standard Deviation, MMSE Mini-Mental State Examination, FMul Fugl-Meyer scale for the upper limbs, FMll Fugl-Meyer Scale for lower limbs, FM Fugl-Meyer Scale, ADM range of motion

Table 3

Characteristics of the study population of cardiorespiratory data



Exercise start (p)

Exercise final (p)

Final recovery (p)


127 + 9

124 + 9 (0.186)

135 + 10 (0.001)

119 + 8 (0.001)*


79 + 5

78 + 4 (0.324)

82 + 6 (0.032)

78 + 6 (0.581)


73 + 10

69 + 9 (0.001)*

114 + 18 (0.001)

73 + 8 (0.874)


20 ± 0.6

18 ± 0.4 (0.001*)

28 ± 0.5 (0.001)

22 ± 1 (0.220)


95 ± 1

95 ± 1 (0.133)

96 ± (0.009)

95 ± 1 (0.829)

SBP systolic blood pressure, mmHg millimeters of mercury, DBP diastolic blood pressure, HR heart rate; bpm beats per minute, RR respiratory rate, rpm repetitions per minute, SAT partial oxygen saturation, % percent, p p value

*p < 0.005 (one-way ANOVA for repeated measures) values expressed as mean ± standard Deviation, values of p refer to the comparison with the “rest” using ANOVA for repeated measures

Time domain indexes showed a significant decrease from R to E, except TINN. The MeanRR decreased at all times (E, RC1, RC2, and RC3) in relation to R (p < 0.001). The SDNN index decreased at all moments (E, RC1, RC2, and RC3) compared to R (p = 0.025, p = 0.001, p < 0.001 and p = 0.043, respectively). There was no statistical difference between the moments RC1 (p = 1.000), RC2 (p = 0.263), and RC3 (p = 0.666) regarding the rMSSD index. The pNN50 index decreased at all moments, but statistical difference was observed only at RC3 (4.9 ± 2 % to 2.2 ± 1 %; p = 0.034). The RRtri index decreased at RC2 (9.1 ± 3 to 7.1 ± 2; p = 0.015). Despite the TINN index was not significant, it showed a downward trend (Fig. 2).
Fig. 2

Analysis of HRV in the time domain at home, aerobic exercise and recovery. *p < 0.005 (one-way ANOVA for repeated measures), the values of p refer to the comparison with the resting (R) using ANOVA for repeated measures; R resting condition, E exercise; RC1 recovery period 1—first 5 min of the recovery period, RC2 recovery period 2—between the 10th and 15th minute of the recovery period, RC3 recovery period 3—between the 25th and 30th minute of the period recovery, MeanRR average RR interval, ms milliseconds, SDNN standard deviation of all normal RR intervals recorded at an interval of time; bpm beats per minute, rMSSD root mean square of the square of differences between adjacent normal RR intervals in an interval of time, pNN50 percentage of adjacent RR intervals with a difference of duration greater than 50 milliseconds, % percentage, RRtri triangular index, TINN triangular interpolation histogram of RR intervals

Figure 3 shows the values of the HRV indexes in the frequency domain obtained in the periods of rest, exercise and recovery. The LF increased at E (394.6 ± 315 vs. 545 ± 82 ms [2], p < 0.001) RC1 (p = 0.001), RC2 (p = 0.004) and RC3 (p = 0.009). The HF reduced R to E (167 ± 130 vs. 11.2 ± 4 ms [2], p < 0.001) and a non significant decrease for the other time points was observed (p = 0.770, p = 0.113, and p = 0.353, respectively). There was no significant change in the LF/HF ration in any moment compared to R.
Fig. 3

Analysis of HRV in the frequency domain at home, aerobic exercise and recovery. *p < 0.005 (one-way ANOVA for repeated measures), the values of p refer to the comparison with the resting (R) using ANOVA for repeated measures; R resting condition, E exercise, RC1 recovery period 1—first 5 min of the recovery period, RC2 recovery period 2—between the 10th and 15th minute of the recovery period, RC3 recovery period 3—between the 25th and 30th minute of the period recovery, LF low frequency component, ms2 milliseconds squared, HF high frequency component, LF/HF ratio ratio between the low-frequency component at high frequency

For quantitative indexes derived from Poincaré plot we measured SD1 (in millisecond); SD2 (in millisecond) and SD1/SD2. There was difference between R and E (16.9 ± 7 ms vs. 5.7 ± 2 ms, p < 0.001), with no statistical difference in other times regarding the SD1 index. There was also difference between R and RC1 (57.2 ± 26 vs.38.6 ± 15 ms, p = 0.003), R and RC2 (57.2 ± 26 vs. 36.26 ± 13 ms, p < 0.001) and a tendency to increase at RC3 (57.2 ± 26 vs. 42.6 ± 18 ms, p = 0.054) with respect to the SD2. We found no significant changes in the SD1/SD2 ratio, except when compared R with E (0.32 ± 0.1 vs. 0.22 ± 0.1 p = 0.027) (Fig. 4).
Fig. 4

Quantitative analysis of the Poincaré plot at resting, aerobic exercise and recovery. *p < 0.005 (one-way ANOVA for repeated measures), the values of p refer to the comparison with the resting (R) using ANOVA for repeated measures; R resting condition, E exercise; RC1 recovery period 1—first 5 min of the recovery period, RC2 recovery period 2—between the 10th and 15th minute of the recovery period, RC3 recovery period 3—between the 25th and 30th minute of the period recovery, SD1 standard deviation of the variability instantaneous beat-to-beat or dispersion of points perpendicular to the line of identity, SD2 standard deviation of long-term continuous RR intervals or dispersion of points along the line of identity, SD1/SD2 ratio scatter of points perpendicular to the dispersion line identity of dots along line identity


Most stroke patients present deficit in gait, among other consequences, an aerobic deconditioning that may compromise their usual activities and provide a higher incidence of cardiovascular events [3643]. The ANS has permanent influence on changes in HR maintaining a role in modulating, moment-by-moment of the cardiovascular system. The stroke has been associated with the onset of dysfunction in the SNA, affecting the proper regulation of HR, this adjustment may be represented by HRV [3, 4].

The analysis of statistical indexes may be obtained by SDNN, pNN50, and rMSSD. The SDNN index, which represents the sympathetic and parasympathetic activity, decreased around 30 min after exercise, showing a decrease in HRV after exercise in patients with stroke. The analysis of HRV in the time domain indexes showed a significant decrease in resting (R) compared to the exercise (E).

Our findings showed a significant decrease in the RRtri at rest (R) compared to exercise (E). The SDNN index decreased at all times compared with rest (R). The pNN50 index decreased only when recovery period 3 (RC3) and RRtri obtained a significant reduction in recovery period 2 (RC2). In the frequency domain the LF index increased from R to E and a significant decrease in the rate compared to recovery moments (RC1, RC2 and RC3) was reported. There was a significant reduction from R to E regarding the HF index. The quantitative analysis of indexes derived from Poincaré graph for SD1 showed only statistical difference between R and E, while there was a change related to the SD2 between RC1 and RC2.

Significant decrease in the SDNN at all times was found compared with the rest (R). The pNN50 index decreased significantly only during the recovery period 3 (RC3). These results are convergent with some authors [4446] who claim that the SDNN decreases in the transition from rest to exercise. Ginsburg et al. [46] conducted a study on stroke patients in the acute phase and found a decreased HRV compared to a group of healthy people of the same age. These researchers found reduced SDNN, RMSSD, and HF in patients with minor stroke.

Conversely, Kouakam et al. [47] analyzed the SDNN, pNN50, RMSSD, and SDANN in 25 stroke patients and found similar values among stroke patients and control group. In this study, there was a high rate of atrial fibrillation (AF) in the group with stroke, but it was not related to the HRV indexes [48]. Mainardi et al. [49] also showed a higher LF component in patients with AF. Some studies [43, 50] showed relationship between oxygen consumption and HR behavior relating to the metabolic demand of the muscle that is being used for exercise, if it is associated with the fact that stroke patients have an energy expenditure greater than the dysfunction motor [51, 52], this may explain the great difference in HRV indexes that reflect sympathetic modulation during exercise. Macko et al. [53], Chu et al. [54], and MacKay-Lyons et al. [55] showed that the VO2 of stroke patients is significantly lower compared with healthy people. Lewis et al. [50] stated that the cardiac output is linearly related to VO2 and the VO2max is greater the greater muscle mass is activated. Thus, the cardiovascular responses to exercise are determined largely by active muscle mass and absolute oxygen consumption, presenting a direct link between systemic oxygen transport and use of this oxygen. To investigate the hypothesis that training on a treadmill improved physical fitness and reduced the energy cost of hemiparetic gait in chronic stroke patients, Macko et al. [53], argue that these improvements may increase functional mobility in stroke patients. In contrast, MacKay-Lyons et al. [55] evaluated the VO2 during maximal treadmill walking at 15 % of body weight support, despite improvements in VO2 training with substantial limitations in exercise capacity.

Teixeira-Salmela et al. [37] showed decreased physical fitness generates a reduction in the practice of daily activities, leading to a circle of reduced physical activity, and has been observed structural changes in the vascular system in the hemiparethic lower limb [40, 53]. Hemiplegic patients have difficulties in maintaining an efficient and comfortable gait speed, around 1.2 meters per second (m/s) requiring greater energy expenditure, which impairs functional mobility.

Bassi et al. [56] argue in their studies that the lower the SDNN values more the unfavorable rehabilitation of patients with ischemic stroke. The lower values of SDNN were independent predictors of an unfavorable outcome functionally. The authors state that an assessment of HRV before a rehabilitation program may provide additional information on the likelihood of functional recovery in this population.

McLaren et al. [57] aimed to determine if autonomic function is impaired after stroke recovery in older patients. The study reported that HRV decreased significantly in elderly patients with stroke. In this study, the overall HRV and baroreflex sensitivity in the low frequency were harmed. HRV appears to be more related to brain injury than the age of the patient.

In our study, the RRTri index also showed a significant decrease compared to rest (R vs. E) and a significant decrease at recovery period 2 (R vs. RC2). The persistent decline of HRV after exercise in patients with stroke is confirmed also by SD2, and RRtri TINN indexes, that represent global cardiac autonomic modulation, also showing that in a period of 30 min the patient does not recover to baseline HRV.

The quantitative analysis of indexes derived from Poincaré graph (SD1, representing the vagal activity) was different between R and E, while the SD2 was different between RC1 and RC2. When we analyzed the indexes that represent parasympathetic activity (RMSSD, pNN50, HF, and SD1) all decreased, but showed no statistical difference between the moment of rest and after exercise moments, except pNN50 in the last 5 min showing the vagal response post exercise. During the exercise, it is expected the fall in these indexes because the sympathetic system is activated, and this was associated with HRV reduction [5861]. Iellamo et al. [59] argue that under physiological conditions a change in vagal activity by changes in blood pressure unlikely contribute to the genesis of oscillations in heart rhythm.

The HR during recovery is associated with parasympathetic reactivation, the return of HR to baseline is associated with good health condition because it is directly linked to physical fitness. Mortality risk is also related to the return of HR. Previous studies showed that when the fall of RHR attenuated mortality risk is major [6266]. Nishime et al. [66] evaluated 9,454 patients who underwent stress testing, the authors reported that both stress electrocardiogram and heart rate recovery were independent predictors of mortality. They suggested that HR recovery may provide additional prognostic information and should be considered for incorporation into routine interpretation of the exercise test. Individuals with better fitness have a higher vagal modulation of HR during exercise and also a better withdrawal of stimulus after exercise [67, 68]. Tullpo et al. [67] evaluated the effects of age and physical fitness on vagal modulation of HR during exercise on a bicycle through the analysis of SD1. The SD1 was higher at rest in younger individuals than individuals in middle-aged or elderly, but age-related differences were smaller during exercise. The authors showed that low physical fitness is associated with impairment of cardiac vagal tone during exercise, while the aging results in greater impairment of vagal function evidenced in rates of home.

Some researches [69, 70] stated that the control mechanisms of HR during exercise is modulated by vagal withdrawal, with no major changes in the adrenergic component, but the recovery seems to be both a decrease in sympathetic activity and increased parasympathetic activity, returning to baseline. There is a disagreement in the literature as to whether the time it takes for the SNA to restore. Imai et al. [70] suggested that vagal reactivation is an important mechanism of slowing heart rate after exercise and that 30 s after cessation of exercise could be a specific index for vagal recovery. Darr et al. [71] concluded that trained individuals, regardless of age, showed a HR recovery significantly faster than untrained individuals. The authors also claim that the slower HR recovery in the elderly may be due to a lack of control variables that influence the recovery. Terziotti et al. [72] showed that after 15 min of recovery, the cardiovascular reflexes were still affected and that vagal stimulation had not completely recovered. After an hour’s rest, there was complete restoration of autonomic control.

Despite in the frequency domain, LF increased from R to E in our study, and a significant decrease compared to recovery times (RC1, RC2, and RC3), and the HF index reduced a from R to E, Dixon et al. [73] stated that there is no change in HF in the transition from rest to exercise in athletes. Casadei et al. [44] claim that the absolute power of the HF band spectrum decreases at the beginning of the exercise, relating this fact to reduce the cardiac vagal activity, the other hand, the normalized power of the high frequency component increases with increasing rates of work. The analysis of LF during all times of the study shows that sympathetic activity does not decrease to the basal levels even after 30 min of rest after exercise and in turn, the HF band, representing parasympathetic activity, remains without significant changes, however, showing a reduced function of the system as it cannot reach the initial levels.

We also found that the LF/HF ratio remained without significant change during the recovery times, considering that this ratio reflects the sympathovagal balance and its stability show us a security in this exercise prescription for a group of stroke patients. On the other hand, Nishioka et al. [74] evaluated the fluctuation of BP in stroke patients during exercise and a negative correlation between the BP and the LF, HF and LF/HF ratio indexes was observed. These authors concluded that lower HRV in patients with stroke may be related to an increase in BP during exercise and this method may be used to estimate the risk in rehabilitation, also suggesting that the assessment of HRV may be useful for exercise prescription in a rehabilitation program.

Xiong et al. [75] and Lakusic et al. [76] reported that the autonomic dysfunction in stroke may persist up to six months after the stroke and parasympathetic dysfunction is prevalent after stroke.

Xiong et al. [77, 78] evaluated patients with ischemic stroke that were divided into two groups, one group of patients who were in the acute phase, another group in the chronic phase and 37 elderly control group. From the analysis of HRV in the frequency domain, stroke patients with acute and chronic presented reduced LF compared to the control group. Patients with acute impairment showed parasympathetic impairment in both tests and those in the chronic phase presented parasympathetic impairment in all tests used compared to the control. The authors stated that autonomic dysfunction occurs in acute ischemic stroke and may persist until six months after the stroke and the parasympathetic dysfunction, rather than sympathetic dysfunction is predominant after ischemic stroke. Importantly, these results refer to the baseline recording and indicatives an injury in postexercise sympathetic cardiovascular adjustments.

Korpelainen et al. [79] conducted a study comparing the RMSSD, LF, HF, and VLF in three stages (acute, 1 month and 6 months after injury) in patients with cerebral ischemia. The results suggested a decrease in HRV, but it tended to disappear more than six months after the injury. The authors of this study also suggested that the fluctuation of LF demonstrates an inadequate response of the baroreflex.

Our study showed that the quantitative analysis of the Poincaré plot can be useful for analysis of autonomic modulation when comparing the exercise period for monitoring the recovery of stroke patients, particularly when analyzing SD1 relating it to the parasympathetic activity. Nevertheless, Mourot et al. [80] showed that in athletes under "overtraining" shows lower values of HF and SD1 in comparison to the resting condition.

Despite the LF index is associated mainly to cardiac sympathetic modulation, Goldstein et al. [81] stated that LF could not be a measure of cardiac sympathetic tone, but rather a measure of baroreflexes. Agreeing with this study, Rahman et al. [82] evaluated the LF index at supine, sitting, and during the Valsalva maneuver, concluding that the LF reflects modulation of cardiac baroreflex and not sympathetic tone.

Some authors [8386] claim that parasympathetic activity is cardiac protective and that people with better cardiopulmonary condition presents better autonomic activity compared to sedentary. Other researches [87, 88] used HRV to identify the intensity of training, determining the balance between the sympathetic and vagal activity, these methods are based on the SD1 index. However, there are no studies on stroke patients attempting to exercise prescription.

Stoller et al. [42] stated that physical exercise in stroke patients is safe even earlier, shortly after the acute event. Traditionally, the rehabilitation of stroke patients is based on recovery of motor function, balance, stretching, and traditional exercises. The study of HRV as well as its help to prescribe exercise may be an important area of research that should be further analyzed in this group of patients, since there were studies [67, 8789] using HRV to identify the intensity of stress, which occurs on the aerobic threshold. It lacks adequate resources to assess the aerobic capacity of patients suffering from severe stroke [42], so perhaps in future work HRV may contribute to one of these methods.

A major difficulty of homogenization of the sample studied is due to the fact that a large majority of stroke patients make use of drugs that alter HRV as beta-blockers. Moreover, the time of 30 min recovery may have influenced the recovery to HRV baseline. Many stroke patients did not support the HR set for individual training due to motor limitations. The absence of a control group comparison of baseline HRV and recovery is also a limiting factor.

Set priorities for stroke prevention requires an understanding of the pattern of disease and exposure to risk factors [90], hence, there is a growing interest in encouraging the practice of physical exercise in patients affected by stroke. HRV is expected to be an important method for the correct prescription of exercise for this group of patients, thus, we conclude that in stroke patients during exercise, the HRV decreases and it does not return to baseline within the first 30 min after exercise, featuring an autonomic imbalance.

Supplementary material

12975_2013_263_MOESM1_ESM.doc (26 kb)
ESM 1(DOC 26 kb)
12975_2013_263_MOESM2_ESM.doc (176 kb)
ESM 2(DOC 176 kb)
12975_2013_263_MOESM3_ESM.doc (35 kb)
ESM 3(DOC 35 kb)

Copyright information

© Springer Science+Business Media New York 2013