Professional Documents
Culture Documents
ABSTRACT INTRODUCTION
T
Mankowski, RT, Michael, S, Rozenberg, R, Stokla, S, Stam, HJ, he second anaerobic threshold (AT2) is frequently
and Praet, SFE. Heart-rate variability threshold as an alternative for used to design and monitor exercise training pro-
spiro-ergometry testing: a validation study. J Strength Cond Res grams for healthy people, athletes, and patients
31(2): 474–479, 2017—Although spiro-ergometry is the estab- with chronic diseases (9,18,30). Specifically, prop-
erly determined AT2 (;90% HRmax) is crucial for estab-
lished “gold standard” for determination of the second ventilatory
lishing appropriate training workloads and intensity needed
threshold (VT2), it is a costly and rather time-consuming method.
to effectively improve exercise performance and to prevent
Previous studies suggest that assessing the second anaerobic
injuries and overtraining (12). However, the “gold standard”
threshold (AT2) on the basis of heart rate variability (HRV) during assessments of AT2 can only be determined through costly,
exercise may be a more cost-effective and noninvasive manner. time-consuming spiro-ergometry testing often in combina-
However, appropriate validation studies, are still lacking. Eleven tion with invasive methods such as multiple blood sampling
healthy, moderately trained subjects underwent 3 incremental for the determination of lactate concentration [La] (5,6).
exercise tests. Ventilation, oxygen uptake (V _ O2), CO2 production Recently, heart-rate variability threshold (HRVT) was
(V_ CO2), and HRV were measured continuously. Exercise testing introduced as a noninvasive, cost-effective, and more
was performed in 3 oxygen (FiO2) conditions of inspired air (14, straightforward method than the currently practiced “gold
21, and 35% of oxygen). Participants and assessors were blinded standard” to assess AT1 and AT2 in an individual (4). The
to the FiO2 conditions. Two research teams assessed VT2s and method has also been proposed to design and monitor indi-
HRVT2s independently from each other. Mean workloads vidualized training programs in athletes and patients under-
corresponding to VT2 and HRVT2 in hypoxia were, respectively, going cardiac and pulmonary rehabilitation (13). However,
appropriate validation and sensitivity analyses to detect
19 6 17% (p = 0.01) and 15 6 15% (p = 0.1) lower in com-
changes in individual training status are still lacking.
parison with hyperoxic conditions. Bland-Altman analysis showed
The HRVT represents a transition point between lower
low estimation bias (2.2%) and acceptably precise 95% limits of
and higher activity of the sympathetic nervous system
agreement for workload 215.7% to 20.1% for all FiO2 condi-
(SNS), which is strongly correlated with an increase in
tions. Bias was the lowest under normoxic conditions (1.1%) in [La], catecholamine concentrations, and minute ventilation
comparison with hypoxia (3.7%) and hyperoxia (4.7%). Heart rate (Ve) that occurs at the AT1 (11).
variability–based AT2 assessment had a most acceptable agree- One of the most commonly used methods for determining
ment with VT2 under normoxic conditions. This simple HRVT2 HRVT is the root mean square of the successive differences
assessment may have potential applications for exercise monitor- (HRVT-RMSSD) (20). Another method is based on the
ing in commercial fitness settings. Poincaré plotting technique, from which the instantaneous
variability of beat-to-beat data is derived by means of disper-
KEY WORDS exercise, maximal testing, cost-effective, healthy sion of points perpendicular to the axis of line-of-identity
subjects, hypoxia, hyperoxia (HRVT-SD1) (26,29).
In healthy subjects, Sales et al. (26) reported strong
Address correspondence to Robert T. Mankowski, r.mankowski@ correlations between the HRVT1 and the “gold standard”
erasmusmc.nl. AT1 assessments (HRVT-RMSSD r = 0.91 and HRVT-SD1
31(2)/474–479 r = 0.93). Thus far, most comparative studies have used
Journal of Strength and Conditioning Research single measurements to investigate agreement between
Ó 2016 National Strength and Conditioning Association these 2 AT determination methods (6,11,26). These
the TM
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
the TM
group-based validation studies, however, do not provide ethics advisory board of Swansea University) and required
information on the sensitivity of HRVT method to detect, subjects to provide informed consent before participation.
e.g., changes in an individual’s AT. Furthermore, there is
evidence that determination of AT2 is more reproducible Experimental Approach to the Problem
than AT1 in trained cyclists (31). Therefore, a comparison Subjects completed 3 incremental exercise tests until exhaustion
of second ventilatory threshold (VT2) and HRVT2 was on the same cycle ergometer (Jaeger ER800; CareFusion,
selected for the purpose of this validation study. To test Houten, The Netherlands, workload accuracy 3%) on
the sensitivity of HRVT for detecting an acute change in different days. Each test was performed under hypoxic
physiological stress, we propose a new validation method (14.04% O2), normoxic (21.2% O2), or hyperoxic (35.08% O2)
based on an acute shift in AT2 by manipulating inspiratory conditions. The tests were separated by at least 72 hours and
oxygen fraction (FiO2). For a given workload, SNS activity performed in a randomized order at the same time of the day.
level will be increased during hypoxic exercise (33) and Randomization occurred by drawing opaque sealed envelopes
lower under hyperoxic conditions (15). In accordance, containing the order of the 3 FiO2 conditions. Both tempera-
AT2 should be reached at a lower workload during hypoxic ture (218 C) and humidity (35–45%) in the laboratory were held
as opposed to a normoxic or hyperoxic exercise stress test. constant. Subjects were asked to refrain from strenuous exercise
To our knowledge, this is the first study to validate the on the day before the test. Furthermore, they were asked to
sensitivity of HRVT2 assessment relative to that of AT2 on refrain from caffeinated drinks at least 12 hours before the
the basis of the breath-by-breath gas analysis method during exercise test.
multiple incremental exercise tests under decreased and
increased FiO2 conditions. Testing Procedures
In accordance, we hypothesized that HRVT2-based Incremental Exercise Test. All the exercise tests were per-
workload is strongly associated with the spiro-ergometry formed at the Clinical Exercise Performance Laboratory
VT2-based workload during incremental exercise test per- from the Erasmus University Medical Center in Rotter-
formed under low, normal, and high FiO2 conditions. In dam, the Netherlands, in the presence of an exercise
addition, we hypothesized that a shift in HRVT2 is associ- physiologist and a sports physician. All subjects were
ated with a shift in VT2. scheduled to perform a symptom-limited exercise stress
test. This standard ramp test started with a 2-minute rest
METHODS period followed by 4 minutes of unloaded cycling as warm-
Subjects up. Next, the workload increased with a slope of 1.8 W/6
A total of 11 (8 males and 3 females, age range 20–28 years) seconds (men) or a 1.2 W/6 seconds (women). Subjects
healthy and medium-trained young adults volunteered to were instructed to cycle until exhaustion with a pedal
participate in this study. The study protocol was approved by frequency of 60–80 rpm. The loaded phase was termi-
the regional Medical Ethics Committee of the Erasmus nated when pedaling frequency dropped below 60 rpm.
University Medical Centre in Rotterdam, the Netherlands A 5-minute unloaded recovery phase ended the exercise
(number: 2012-128 [NTR3777]). All participants provided stress test. During the warm-up, workload, and recovery
written informed consent. The study conforms to the Code phases of the exercise protocol, V_ O2, V_ CO2, minute venti-
of Ethics of the World Medical Association (approved by the lation (Ve) (Oxycon Pro; Carefusion; accuracy: volume
sensor 2%, O2 and CO2 ana-
lyzers 0.5%, CO2) were moni-
tored. Peak oxygen uptake,
TABLE 1. Subjects’ characteristics.* workload, heart rate (HR), and
respiratory exchange ratio
Hypoxia Normoxia Hyperoxia
(RER) were documented.
(N = 11), (N = 11), (N = 9),
N = 11 mean 6 SD mean 6 SD mean 6 SD Immediately after each exer-
cise, test subjects were asked
Age (yrs) 23.6 6 2.2 to rate their perceived exertion
Weight (kg) 69.1 6 9.8 on a scale from 6 to 20 using
Height (cm) 177.9 6 8.2
22 the Borg scale (1).
BMI (kg$m ) 22.0 6 1.7
V_ O2peak (ml$kg21$min21) 46.8 6 5.3 54.6 6 7.0 54.2 6 6.7
Wmax (W) 289 6 48 324 6 54 336 6 55 Heart Rate and Heart Rate Vari-
Borg score 16.4 6 0.7 16.3 6 0.8 16.2 6 0.8 ability. Beat-to-beat HR was
measured continuously using an
*The results in boldface (hypoxia) were significantly different from normoxia and hyperoxia
(p # 0.05). HR monitor (Zephyr BioHar-
ness 3; Zephyr, Annapolis, MD,
USA, HR accuracy 63 bpm).
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
HRV & VT
Data analysis
Determination of the Second Variability Threshold. Obtained
values of V_ O2, V_ CO2, RER, and (Ve) were averaged for
each 30-second time window. Second ventilatory thresh-
old was determined independently by 2 sports physicians,
both having more than 5 years of experience with clinical
assessment of cardiopulmonary exercise tests. All tests
results were encoded and both assessors were blinded
to the 3 different FiO2 test conditions. For the present
study, VT2 was defined by the second nonlinear increase
of the Ve/V_ CO2 curve (24). When VT2s differed between
the 2 assessors, consensus on VT2 was obtained by re- Figure 1. Bland-Altman scatter plot. Mean workload values of second
variability threshold (VT2) and the second heart-rate variability threshold
viewing each other’s assessment. Ventilatory threshold (HRVT2) are plotted on X axis. Empty circles—hypoxia; half-filled circles—
was expressed as workload (W) level corresponding normoxia; filled circles—hyperoxia; LoA = limit of agreement. Heart rate
with VT2. variability (HRV) was quantified every 30 seconds by first calculating the
root mean square of successive differences of R-R intervals (RMSSD),
then dividing this by the average R-R interval over the corresponding 30-
Determination of Second Heart Rate Variability Threshold. The second period, to calculate the normalized RMSSD (nRMSSD).
second heart-rate variability threshold was determined by 2
the TM
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
the TM
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
HRV & VT
improvement in training status. In particular, they observed HRVT may benefit from the application of objective break-
similarly increased workload corresponding to VT2 point determination methods.
(;17.5%) and HRVT2 (;15%). The present study extends The current study has some limitations. In particular,
on their results by showing a rather similar difference in regarding the identification of HRVT2, a clear upward
workload capacity for VT2 when manipulating FiO2 con- inflection point could not always be identified. Therefore, 2
ditions. Higher FiO2 (hyperoxia) decreases and lower FiO2 measurements were excluded, as both the 2 independent
(hypoxia) increases the activity of SNS during exercise assessors could not detect an inflection point corresponding
(15,33). In comparison with a medium- or long-term train- to HRVT2.
ing intervention study, our approach was intended as a more The currently used, and as far we know, best available
direct method to test the sensitivity of the HRVT assess- VT and HRV methods both require a visual assessment of
ment method in response to a change of physiological inflection points in both ventilation and HRV data. As
stress. The shift in VT2 under hypoxic and hyperoxic con- there is no gold standard for determining VT, either of the
ditions is well in line with previous reports (27). In particu- methods may have introduced some level of bias or error.
lar, supplementary oxygen may also improve oxygen Based on the results of this study, it is suggested that
delivery and consumption, and hence exercise tolerance HRVT2 assessment method may give valid results under
(V_ O2max). Therefore, time to exhaustion is extended and normoxic conditions and not under hypoxic or hyperoxic
activation of SNS is lower during a maximal test under conditions. Although all exercise tests were performed
hyperoxic conditions in comparison with hypoxic condi- according to exercise testing guidelines, the equipment
tions (23,25). The HRVT2 method seems less sensitive as required to perform a precise ramp test on a cycle
the shift in HRVT2 did not reach statistical significance. ergometer at 1.8 and 1.2 W$s21 is not trivial. In accor-
The latter result may also be explained by a type II error dance, future HRV-VT validation studies should preferably
as we had to exclude 2 HRVT measurements under hyper- also include an exercise testing protocol using a graded
oxic conditions because of difficulties to determine a clear step test (30 seconds to 3 minutes at a constant workload)
inflection point. The lack of statistical difference between on a commercially available cycle ergometer under nor-
normoxia and hypoxia may also be caused by limited sta- moxic conditions.
tistical power in our study. Another explanation, in line with Although computationally simple and achieving a high
the high SDs of our measurements is the recently described temporal resolution, time-domain HRV measures such as
variation in the individual response to hypoxia/hyperoxia nRMSSD have not previously been validated for determin-
(8,16,19). Therefore, HRVT2 assessments may be valid for ing VT2. Previously, frequency-domain measures such as the
tests under normoxic conditions. product of HF power and HF frequency (3,4) have been
In the present study, an upward inflection in the time- used. Future research could be directed toward refining the
domain HRV measure nRMSSD (RMSSD:RRI ratio) was application of time-domain or alternative HRV analysis to
used to determine HRVT2 as opposed to frequency domain estimate ventilatory thresholds.
measures (3,4). By comparison, frequency domain measures In conclusion, the present study showed that there is
should not be applied to a time window of less than 10 times a most acceptably precise agreement between the VT2
the lower bound of the frequency band (i.e., 67 seconds for and HRVT2 assessment methods under normoxic con-
a high-frequency lower bound of 0.15 Hz) (10). ditions. This simple HRVT2 assessment may have poten-
Therefore, the present study used the time-domain tial applications for exercise monitoring in commercial
measure nRMSSD (RMSSD:RRI ratio), using 30-second fitness settings.
windows, which was perhaps the simplest to perform (10).
Heart rate variability measurements are sensitive to the PRACTICAL APPLICATIONS
autonomic nervous system disturbances in pathologies (e.g., These results open up new avenues for a cost-efficient and
Parkinson’s disease, diabetes-related autonomic neuropathy) noninvasive HRVT method that could be easily applied to
(32) and to exercise performance improvement. Especially, determine, monitor, and individualize the appropriate work-
during exercise, the HRV signal can also be influenced by load intensity in different types of aerobic exercise training
nonneural factors (17). In particular, the nonneural influences modalities in commercial fitness settings.
of respiration on the sino-atrial node, particularly during
exercise at and above VT2, likely contribute to the upward ACKNOWLEDGMENTS
inflection of nRMSSD (10). This makes the time-domain We acknowledge Mr. Bert Bannink and the pulmonology
method practically useful as a proxy measure for VT2 detec- department at the Erasmus University Medical Centre for
tion. The present study used visual determination of break- access to measuring devices and constant technical support.
points for HRVT and VT. Although this approach is widely We thank Mr. Oskar de Kuijer from Vital B.V. for the HRV
used in a range of physiological contexts, mathematical measuring equipment and software. Finally, this study
models have been shown to be useful to objectively identify would not have been possible without our flexible and
threshold breakpoints (2). Accordingly, future studies on committed subjects.
the TM
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
the TM
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.