Content uploaded by Sara R. Sherman
Author content
All content in this area was uploaded by Sara R. Sherman on Feb 11, 2021
Content may be subject to copyright.
Heart-Rate Variability Recording Time and Performance
in Collegiate Female Rowers
Sara R. Sherman, Clifton J. Holmes, Bjoern Hornikel, Hayley V. MacDonald,
Michael V. Fedewa, and Michael R. Esco
Purpose:To assess the agreement of the root mean square of successive R-R interval (RMSSD) values when recorded
immediately upon waking to values recorded later in the morning prior to practice, and to determine the associations of the
RMSSD recordings with performance outcomes in female rowers. Methods:A total of 31 National Collegiate Athletic
Association Division I rowers were monitored for 6 consecutive days. Two seated RMSSD measurements were obtained on at
least 3 mornings using a smartphone-based photoplethysmography application. Each 1-minute RMSSD measure was recorded
following a 1-minute stabilization period. The first (T1) measurement occurred at the athlete’s home following waking, while the
second (T2) transpired upon arrival at the team’s boathouse immediately before practice. From the measures, the RMSSD mean
and coefficient of variation were calculated. Two objective performance assessments were conducted on an indoor rowing
ergometer on separate days: 2000-m time trial and distance covered in 30 minutes. Interteam rank was determined by the coaches,
based on subjective and objective performance markers. Results:The RMSSD mean (intraclass correlation coefficient = .82;
95% CI, .63 to .92) and RMSSD coefficient of variation (intraclass correlation coefficient = .75; 95% CI, .48 to .88) were strongly
correlated at T1 and T2, P<.001. The RMSSD mean at T1 and T2 was moderately associated with athlete rank (r=−.55 and
r=−.46, respectively), 30-minute distance (r= .40 and r= .41, respectively), and 2000 m at T1 (r=−.37), P<.05. No significant
correlations were observed for the RMSSD coefficient of variation. Conclusion:Ultrashort RMSSD measurements taken
immediately upon waking show very strong agreement with those taken later in the morning, at the practice facility. Future
research should more thoroughly investigate the relationship between specific performance indices and the RMSSD mean and
coefficient of variation for female collegiate rowers.
Keywords:athletic monitoring, female athletes, RMSSD, rowing, HRV recording time
In recent years, sport-science technology has been at the
forefront of athletic monitoring and performance across numerous
sport disciplines.
1
Among these technologies are the use of short-
term measurements of heart-rate variability (HRV), which reflects
cardiac autonomic modulation of the heart by expressing the non-
linear and complex oscillations that occur between successive heart
beats.
1,2
HRV monitoring has gained popularity among coaches
and athletes as a simple, noninvasive tool to monitor autonomic
function for performance and recovery.
1,3
Several studies have
shown that, during intense training periods, vagal indices of HRV
decrease acutely and rebound 24 to 48 hours beyond their pretraining
recovery levels, resulting in different performance outcomes for both
recreational and highly trained athletes.
4–7
Endurance-type sports,
such as rowing, place excessive demands on both the aerobic and
anaerobic systems.
8,9
Such demands result in major physiological
perturbations, often characterized by significant fluctuations in
heart rate (HR) and, thus, autonomic function.
5,6
Given the unique
physiologic profile of rowers
8
and their typical high training vo-
lumes, monitoring day-to-day fluctuations in HRV could provide
valuable information regarding recovery and adaptation to training.
The time-domain HRV parameter, root mean square of succes-
sive R-R intervals (RMSSD), is the preferred marker of parasympa-
thetic modulated HRV in ambulatory or field-based conditions
among athletic populations
10,11
in resting conditions across a weekly
(or rolling) average. RMSSD is often used with athletes because it is
not significantly influenced by breathing frequency,
12
can be cap-
tured over a short time frame,
13
and is easily calculated in Microsoft
Excel (Microsoft Corp, Redmond, WA).
14
Nonetheless, time con-
straints in a collegiate sport environment (ie, training scheduling,
compliance with National Collegiate Athletic Association [NCAA]
regulations, etc) and irregular practice hours are typical challenges
faced when implementing monitoring routines in distinct sports
settings. To circumvent these issues, several physiologic monitoring
devices, including those that measure resting HRV, have been
adapted for use in nonlaboratory settings. However, HRV metrics
(ie, RMSSD) are extremely sensitive to both methodological and
physiological perturbations.
2
Therefore, HRV measurements upon
waking are preferred to limit the potential influence of external
stimuli on the autonomic nervous system and, thus, cardiac auto-
nomic function measured via RMSSD.
1,11,15,16
The feasibility of
obtaining high athlete compliance and reliable RMSSD measures in
an unsupervised, home-based setting remains a concern among
coaches and researchers alike. Morning resting RMSSD measures
performed on-site at the beginning of regularly scheduled practice
sessions, with oversight from trained technicians, seem to represent
the most practical and accurate monitoring solution for collegiate
athletics. Yet, based on the existing literature, which presents
different times of day to perform ultra-short-term RMSSD collec-
tion,
2
it is unclear whether the reliability of short-term RMSSD
measurements is influenced by the timing of when they are per-
formed in the morning hours (ie, measured immediately upon
waking vs immediately prior to morning practice). Considering
The authors are with the Dept of Kinesiology, University of Alabama, Tuscaloosa,
AL, USA. Sherman is also with the Integrative Physiology Laboratory, College of
Applied Health Sciences, University of Illinois at Chicago, Chicago, IL, USA.
Sherman (ssherm5@uic.edu) is corresponding author.
1
International Journal of Sports Physiology and Performance, (Ahead of Print)
https://doi.org/10.1123/ijspp.2019-0587
© 2021 Human Kinetics, Inc. ORIGINAL INVESTIGATION
that research collected at different time points may involve different
autonomic-nervous-system-altering stimuli, it is uncertain whether
researchers are deriving accurate and reliable values of RMSSD and
their respective daily fluctuations.
To our knowledge, there are no existing studies that have
investigated the within-and between-day reliability of short-term
HRV measurements among athletes and the association with athletic
performance. Therefore, the primary aim of this study was to assess
the agreement between HRV values that were recorded immediately
upon waking with the values recorded later in the morning, upon
arriving at the practice facility, prior to practice. In addition, this study
sought to determine the associations of these HRV measures with
performance outcomes in competitive female rowers.
Methods
Participants
A total of 39 females, NCAA Division I collegiate rowers were
recruited from the University of Alabama(Tuscaloosa, AL) women’s
rowing team to participate in this study. The athletes were members
of the same varsity team, with 4 (3) years of competitive rowing
experience. Each of the athletes passed a medical examination from
university physicians prior to participation in rowing-related activi-
ties, and pertinent health histories were collected from the team’s
medical staff. Specific exclusion criteria included the following:
(1) any diagnosed disability or disease that could influence response
to training and/or HR, (2) an injury not allowing full participation in
the performance tests, (3) any medication effecting HR, or (4) asthma
or any respiratory/breathing disorder. Menstrual cycle and contra-
ception use were identified for each of the participants, but with little
conclusive evidence toward this effect on vagal tone, these partici-
pants were not excluded from analysis.
17,18
Finally, the participants
were asked to refrain from alcohol consumption for the entirety of the
study, and all subjects self-reported this abstinence.
Written informed consent was obtained by each participant
after learning of any potential risks associated with their involve-
ment in the study. This study was approved by the University
of Alabama Institutional Review Board in accordance with the
Declaration of Helsinki, as well as NCAA Compliance personnel.
Overview of the Study Design
The data collection for this observational study occurred over
1 week during the beginning phase of the participants’winter
training program (January 2018). A 2-week familiarization period
preceding the 6-day data collection period, where the athletes were
introduced to the HRV recording method, is described below.
HRV Recordings
The HRV recordings were obtained using a previously validated
19
photoplethysmography smartphone application, HRV4Training
(https://www.hrv4training.com/), downloaded onto each athlete’s
personal mobile device. For all recordings, the athletes placed their
left index finger directly on the posterior camera sensor of their
mobile device. The RMSSD recordings were performed in the
upright seated position, with their back supported comfortably by
a stable, backed chair, in order to reduce potential parasympathetic
saturation commonly observed among highly fit individuals
with low resting HR.
20
The athletes were instructed to limit any
bodily movement and practice spontaneous breathing
11,19
before
opening the HRV4Training application. The rowers then initiated a
1-minute stabilization period, followed by a 1-minute data acqui-
sition period.
19
Ultra-shortened (ie, 2-min) measures of RMSSD
have been validated to be accurate and reliable when compared
with the 10-minute criterion measure recommended by the task
force,
2
which is impractical in such athletic environments.
15,21,22
During the week of data collection, the rowers were asked
to complete 2 ultra-shortened morning RMSSD measurements,
using the same personal mobile device with the HRV4Training
photoplethysmography application and procedure as in the 2-week
familiarization period. However, unlike during the familiarization
period where only 1 RMSSD measurement was recorded per day
(for a total of 14 d), the study week involved a second RMSSD
measurement. The first HRV measurement (T1) was performed at
the athlete’s home immediately after waking and elimination, in a
seated position with a backed chair. They then proceeded through
their normal, prepractice morning routine prior to their arrival at
the boathouse. The second HRV measurement (T2) was obtained
within 1 hour of T1, upon arrival at the on-campus boathouse for
daily practice, according to the same procedures used in T1. All T2
measurements were collected between 5:30 and 6:00 AM in a quiet
area of the boathouse, away from the main boat bay, under the
supervision of trained graduate research assistants (Figure 1).
Training Program
The same rowing coach provided the training plan for all rowing
athletes; as such, the content was similar across all rowing athletes and
consisted of on-the-water rowing practice, land-based ergometer
training, and strength training. The individual athlete’s rowing ergom-
eter and on-water performance were collected from the coaches at the
onset of the data collection and continued throughout the entire study
period. A training program, including recording times, is outlined in
Figure 1and Table 1. The athletes did not compete in any formal races
during the week of data collection, nor in the period leading up to it.
Performance Metrics
Two land-based rowing ergometer tests (2000-m time trial and
distance covered in 30 min) were included as performance markers
to accompany the third metric, team ranking. The performance
records included split and time to completion for each of the athletes’
2000-m time trial, with a shorter time to completion indicating a
greater performance. The 2000-m time trial is a common perfor-
mance metric used in the sport of rowing to objectively quantify a
rower’s ability to effectively move their mass through the stroke
compared with their teammates’and competitors’ability.
9
Success-
ful performance requires an excess of anaerobic threshold, often
resulting in ≥95%
˙
VO2max.
8
Similarly, the 30-minute ergometer
test reflects the number of meters pulled on anindoor ergometer over
the course of 30 minutes. The typical NCAA rowing season begins
in September, with longer race lengths (ie, 4000–10,000 m) de-
signed to build an endurance base with technical proficiency for
the more anaerobically taxing spring races, generally composed of
2000-m tests.
8,9
Thus, the winter training period (November–March)
of the season focuses on the shift from primarily aerobic to anaerobic
thresholds. The rank of each rower within the team (54 athletes in
total) was formulated at the end of each week at the discretion of
the coaching staff. This ranking system was based on a culmination
of the following factors: (1) differences in scores in biweekly
ergometer testing, which occurred on Monday and Friday after-
noons; the 2 ergometer tests that occurred in the week of data
collection were the 2000-m time trial (Monday) and the 30-minute
(Ahead of Print)
2Sherman et al
meters (Friday) (see Table 1for further reference to training intensity
and type); (2) racing performance, including but not limited to time
trials, small and/or large boat matrices, and/or seat racing; (3) factors
relevant to crew combination, namely within crew compatibility
individual coachability, technical compatibility, and team balance/
harmony; and (4) assessment of competitive readiness and other
factors relevant to achieving team objectives (ie, academic eligibil-
ity, injury, etc).
23
Data and Statistical Analyses
The existing literature
24,25
suggests that training adaptation is more
precisely reflected by weekly mean values of HRV than by isolated
recordings due to the daily perturbations characteristic of auto-
nomic nervous system activity and, thus, HRV. Furthermore, the
RMSSD weekly average consisting of 3 or more measures per
week or a 7-day running average has been shown to accurately
identify any vagal-related changes in athletes and is considered
Figure 1 —Training program, including 2 HRV recording times. HRV indicates heart-rate variability; T1, first measurement; T2, second
measurement.
Table 1 Full Daily Workout Routine Completed by Each Rowing Athlete Throughout the Study
Day Daily workouts
Monday Aerobic warm-up run: 12′
Water workout AM: 24′outbound, 5–7′recovery, 30′return; both at 18/20 spm, alternating every 2c
Rowing ergometer workout PM: 2000-m time trial
Tuesday Aerobic warm-up run: 12′
Water workout AM: 4 ×10′w/3.5′recovery; at 16/18/20/22 spm, alternating every 4′/3′/2′/1′
Strength workout AM: goal = approximately 60% 1RM—box jumps (3 ×5); front squat (2 ×6); back squat (3 ×8); low box step-up
(3 ×5); RDLs (3 ×6)
Wednesday Aerobic warm-up run: 12′
Water workout AM: 4 ×1.5 m w/5′recovery; (format: 1000 m at 24–26 spm, 500 m at 30 spm rate cap, but must drop split). Each piece
with static scrimmage start (1 stroke and go).
Thursday Aerobic warm-up run: 22′
Water workout AM: 10 ×(3′on/1′off) w/5′–7′recovery between piece 5 and 6; pieces 1–5 at 24 spm and pieces 6–10 at 26 spm
Strength workout AM: goal = approximately 60%–70% 1RM—clean pull (3 ×5 and 2 ×3); scalp retractions (3 ×5); bench press (3 ×6);
DB incline bench (3 ×5); DB row (3 ×5); upright row (3 ×5)
Friday Aerobic warm-up run: 12′
Water workout AM: 60′drill; 30′–40′SS row; both at 20–24 spm
Rowing ergometer workout PM: 30-min ergometer test
Saturday Aerobic warm-up run: 12′
Water workout AM: 60′of drill/steady state row at 20–24 spm; 2 ×2′w w/equal recovery (format: 1500 m at 26 spm; 500-m drop split,
rate cap at 30 spm)
Abbreviations: 1RM, 1-repetition maximum; cap, maximum stroke rate possible per minute; DB, dumbbell; piece, specific interval during the workout; RDL, Romanian
deadlift; split, amount of time it would take to complete 500 m; spm, strokes per minute; w/, with.
(Ahead of Print)
HRV Recording Time and Performance in Female Rowers 3
a criterion measure for tracking weekly changes in autonomic
function in response to training.
25
Therefore, only those athletes
who completed both measurements (T1 and T2) on at least 3 of the
6 recording days were included in the final analysis.
Analyses were performed using SPSS (version 23.0; IBM Corp,
Armonk, NY) and Microsoft Excel 2016 software. The RMSSD
mean and coefficient of variation (CV) were calculated (CV = [SD/
mean] ×100; %) for each of the 2 daily recordings on an Excel
spreadsheet. The RMSSD coefficients of variation values represent
the standard error of the estimate (SEE; ie, absolute reliability) within
the interday RMSSD assessments. The mean and CV RMSSD
values collected for each athlete at both time points were compared
using paired samples ttests. A Shapiro–Wilk test was used to assess
the normality of the studied variables and was log transformed in
the case of nonnormality. Thus, all daily RMSSD values were log
transformed (mean [LnRMSSD
M
]andcoefficient of variation
[LnRMSSD
CV
]), and these values were used for all analyses.
1,3,11
The relative reproducibility of the LnRMSSD
M
and
LnRMSSD
CV
measures between T1 and T2 was calculated using
Hopkins spreadsheets.
26
The intraclass correlation coefficient and
the typical error of measurement were calculated with 95% confi-
dence intervals for reliability determination. The total distribution
of the RMSSD mean and RMSSD coefficient of variation (ie, SEE,
bias ± 95% limits of agreement [LOA]) scores from the 2 mea-
surement points are presented in Bland–Altman plots.
27
The 95%
LOA for the total T1 and T2 distributions were generated by
multiplying the SD of the score difference by 1.96, then ± from the
mean difference to create the upper and lower limits, respectively.
Pearson rcorrelational analyses were performed to determine if
the HRV parameters (LnRMSSD
M
and LnRMSSD
CV
)wererelated
to the 2000-m time or 30-minute meters performance tests. Spear-
man Rho correlational analyses were performed to determine the
relationship between rank at T1 and T2 for LnRMSSD
M
and
LnRMSSD
CV
. Hopkins thresholds were used to interpret the mag-
nitude or strength of the correlation coefficients (ie, intraclass
correlation coefficient, Pearson r, Spearman Rho) as follows:
<.10 (trivial), .30 (small), .50 (moderate), .70 (large), .90 (very
large), and 1 (nearly perfect).
26
A priori alpha was set to P<.05, and
all data are presented as mean (SD).
Results
A total of 31 (n = 31) athletes were included in the final analyses
(19.7 [1] y, 26.3 [4] kg/m
2
). Six athletes were excluded for lack of
adequate compliance (ie, at least 3 recordings during the week and
2 each day), and 2 athletes sustained injuries during the study,
which did not allow them to complete the assigned workouts.
Two subjects were not included in the 30-minute ergometer test
correlation because their 30-minute test took place at a different
time than the rest of the group due to their academic schedules.
The mean of the differences between daily recordings was 34
(13) minutes.
The 2000-m time was 459 (16) seconds (n = 31); the 30-minute
distance ergometer test was 7143 (235) m (n = 29); and the rank
ranged between 1 and 54, as there were 54 athletes on the team. Very
large intraclass correlation coefficients were observed between T1
andT2forLnRMSSD
M
and LnRMSSD
CV
(P<.001), respectively
(Table 2). Bland–Altman plots showed tight LOA for both the
LnRMSSD
M
and LnRMSSD
CV
(Table 2; Figures 2and 3). The
mean HR for T1 was 67 (9) beats per minute and 68 (11) beats per
minute for T2, as recorded in the same window as the HRV values.
The LnRMSSD
M
values obtained at both time points were moder-
ately correlated with all of the performance indices (P<.05), except
the 2000-m test at T2 (r=−.33, P= .70, small). The LnRMSSD
CV
was not significantly correlated with any of the performance measures
at T1 (rank: P= .124, small; 2000-m time: P= .08, moderate; 30-min
meters: P= .052, moderate), or T2 (rank: P= .57, trivial; 2000-m
time: P= .29, small; 30-min meters: P= .44, small) (Table 3).
Discussion
The purpose of this study was 2-fold, as follows: (1) to compare the
RMSSD values recorded immediately upon waking to the values
recorded later in the morning, prior to practice and (2) to determine
the associations of these measures with performance outcomes in
competitive female rowers. To our knowledge, this is the first study
to investigate the agreement of ultra-short LnRMSSD measures
spanning an entire week among Division I collegiate female
rowers, which represents an understudied population.
Interestingly, the CV of daily log-transformed RMSSD
(LnRMSSD) values, which generally characterize perturbations
to cardiac autonomic homeostasis (ie, the day-to-day fluctuations
of HRV),
10
were not correlated with any performance indices at
either time point, despite a strong agreement between the T1 and T2
measurements. These findings may speak more to the limitations
of the current correlational study design rather than the lack of
importance of monitoring LnRMSSD
CV
longitudinally. Specifi-
cally, without consideration of the effects of the overall training
load performed by the athletes during the week of data collection,
it may be difficult to detect meaningful relationships to specific
performance outcomes. Furthermore, this study was completely
observational, where the training plan was not altered by the
researchers to induce specific perturbations to homeostasis and,
as such, did not result in large fluctuations in LnRMSSD
CV
.It
Table 2 Comparison of the LnRMSSD
M
and LnRMSSD
CV
Between Home and Boathouse Measurements
Limits of agreement
HRV
variable
Recording
location
Mean
(SD)
ICC
(95% CI) SEE TE
Bias
(95% CI) 1.96*SD
Lower limits
(95% CI)
Upper limits
(95% CI)
LnRMSSD
M
T1 4.5 (0.5) .75 (.48 to
.88)*
0.29 0.40 −0.03 (−0.18 to
0.12)
0.80 −0.83 (−1.09 to
−0.57)
0.76 (0.51 to
1.02)
T2 4.5 (0.5)
LnRMSSD
CV
T1 8.7 (3.7) .82 (.63 to
.92)*
2.73 3.48 0.35 (−0.94 to
1.64)
6.91 −6.56 (−8.80 to
−4.32)
7.26 (5.02 to
.9.50)
T2 8.4 (4.4)
Abbreviations: CI, confidence interval; HRV, heart-rate variability; ICC, intraclass correlation coefficient; LnRMSSD
M
, RMSSD log-transformed mean values;
LnRMSSD
CV
, RMSSD log-transformed coefficient of variation; RMSSD, root mean square of successive R-R intervals; SEE, standard error of the estimate; T1,
home (1st) measurement; T2, boathouse (2nd) measurement; TE, technical error.
*p<0.001
(Ahead of Print)
4Sherman et al
could be that the training program (Table 1) was not sufficient to
evoke patterns of variability, or the rowers were already well
adapted to the given training plan, as the period of data collection
occurred in the middle of the season, which could thus explain
the lack of changes in LnRMSSD
CV
that have been detected in
previous studies.
10,21,28
For example, Flatt and Esco
28
found that
weekly LnRMSSD
CV
values were more sensitive to training load
adjustments than LnRMSSD
M
values across 3 weeks of varying
training load in female collegiate soccer players. Finally, it is
important to note that the performance markers that were chosen
for this study were also not specifically altered or assigned by
the research staff themselves, and while the 2000-meter time trial,
30-minute meters, and rank are common markers of performance
Figure 2 —Bland–Altman method comparing the differences of the means for T1 and T2 LnRMSSD
M
measurements. The middle line represents the
mean bias between the home (T1) and boathouse (T2) measurements (−0.03). The upper limit of agreement is represented by the top line (0.76), and the
lower limit of agreement is represented by the bottom line (−0.83), as seen in Table 2. Circles represent individual data points. LnRMSSD
M
indicates
RMSSD log-transformed mean values; RMSSD, root mean square of successive R-R intervals; T1, first measurement; T2, second measurement.
Figure 3 —Bland–Altman method comparing the differences of the means for T1 and T2 LnRMSSD
CV
measurements. The middle line represents the
mean bias between the home (T1) and boathouse (T2) measurements (0.35). The upper limit of agreement is represented by the top line (7.26), and the
lower limit of agreement is represented by the bottom line (−6.56), as seen in Table 2. Circles represent individual data points. LnRMSSD
CV
indicates
RMSSD log-transformed coefficient of variation; RMSSD, root mean square of successive R-R intervals; T1, first measurement; T2, second
measurement.
Table 3 Pearson rfor LnRMSSD
M
and LnRMSSD
CV
and Performance Variables
LnRMSSD
M
LnRMSSD
CV
Performance variable T1 T2 T1 T2
Rank −.55* −.46* −.25 −.06
2000 m −.37* −.33 −.32 −.2
30 min
‡
.40* .41* .37 .15
Abbreviations: CV, coefficient of variation; LnRMSSD
CV
, RMSSD log-
transformed CV; LnRMSSD
M
, RMSSD log-transformed mean values; RMSSD,
root mean square of successive R-R intervals; T1, home (1st) measurement; T2,
boathouse (2nd) measurement.
*p<0.001; ‡n = 29, all other measures n = 31.
(Ahead of Print)
HRV Recording Time and Performance in Female Rowers 5
in collegiate rowing settings,
9,23
they may not have been the most
appropriate for detecting changes in LnRMSSD
M
or CV for this
population. Future studies should investigate if ultra-shortened
LnRMSSD
CV
measures are a reliable marker of training adaptation
and performance indices in collegiate female athletes throughout
various periods of their competitive seasons.
One other study has investigated the reliability of ultra-shorted
HRV measurements across and between days in team sports.
Nakamura et al
24
demonstrated superior interday and intraday
reliability between LnRMSSD measures in elite rugby union players
between 4 consecutive days during a team’s training camp. How-
ever, this study only investigated intraday reliability on the first
training day of the week, and both of the recordings were conducted
at the practice facility within only 10 minutes of the other. Thus,
it would be difficult to compare the results of the current study,
which investigates a weekly rolling LnRMSSD average, to that of
only 1 intraday reliability measure. The weekly rolling average of
LnRMSSD collected in the field from at least 3 time points across
7 days does not exactly parallel research models,
25
which suggests
that 3 random time points across 7 days is sufficient for a true weekly
rolling average. However, a minimum of 3 time points does align
with standard practice by coaches with moderate to large teams, such
as the current sample. While some may consider this a limitation, it
seems the most practical alternative to artificially eliminating impor-
tant data to meet a minimum standard of collected days, which could
potentially lead to assumptions of data manipulation. The overarch-
ing purpose of the aforementioned study, along with the current
study, is to make the HRV measurement more accessible for
monitoring training adaptations in the field while considering the
actual situations for which the recordings would take place.
Though the sex of the rugby players is not specifically stated
in Nakamura et al’s
24
study, the majority of the current HRV
literature has been conducted in athletic populations consisting of
elite (ie, Olympic or national team level), male athletes
3,7,28
and,
thus, may not be generalizable to the current study sample or other
studies involving female athletes. It is unclear how this may have
influenced our results, if at all. Future research should consider the
effects of variables specific to female athletes in both short- and
long-term study designs and their interactions with the sensitivity
of HRV measures in parallel with individual weekly training
adaptation.
Practical Application
This study utilized a unique approach to HRV monitoring, as it was
designed to represent the “real-life”daily activities of the rowers,
with limited interruption. Given the tendency of coaches and
practitioners to gather HRV data in the practice facility
15,21,22
in order to overcome compliance (ie, in this study only 31 of
the 54 total rowers volunteered and completed the current study to
its length) and financial burdens (ie, ordering only one HRV
recording tool to be shared vs ordering enough for a full team),
these results are promising for the future of sports science and
athletic monitoring. These novel findings suggest that, despite the
complexity of cardiac autonomic modulation (captured via HRV
measurements)
29,30
and its sensitivity to both internal and external
stimuli, particularly outside of a controlled laboratory setting, both
LnRMSSD
M
and LnRMSSD
CV
, when recorded at the practice
facility, tend to have tight LOA with the values produced at home,
upon waking. Furthermore, trends in LnRMSSD
M
, when compared
with performance measures, can be detected with moderate corre-
lations, with the largest correlations detected in the morning
measures. Readers should be aware of the potential decreased
sensitivity of measures taken later in the morning when attempting
to monitor performance readiness (ie, rank performance measure:
r=−.55 at T1 and r=−.46 at T2; see Table 3). Furthermore,
the lack of correlation of performance with LnRMSSD
CV
values
should not be attributed to the variable itself, but the current study
design and correlations of this nature warrant further research. Our
data may lead these practitioners, coaches, and trainers to ponder
if this decreased sensitivity is more important than overall compli-
ance and absolute accuracy of the measurements in terms of
effective cardiac autonomic athletic monitoring.
Conclusion
Our findings address an important methodological question regard-
ing the time point at which HRV measurements should be recorded.
These data suggest that daily monitoring of the LnRMSSD
M
and
LnRMSSD
CV
measurements performed at home, upon waking in a
fasted state (T1), show very strong agreement with those taken later
in the morning, prior to practice at the rowing facility (T2).
Furthermore, measures of LnRMSSD
M
were moderately correlated
to rank and the 30-minute test at both T1 and T2, and the 2000-m
test only for T1. No significant correlations were detected between
any of the performance measures and LnRMSSD
CV
for either
T1 or T2.
Acknowledgments
Thank you to the University of Alabama NCAA DI Women’sRowingTeam
athletes, coaches, and training staff for allowing the success of this study. Your
time and dedication to this science and sport, which is the first of any female
collegiate rowing team of your caliber, will undoubtedly help to inspire other
young scientists to narrow the gap in female athletics and research. Thank you
to Dr Todd Freeborn for your willingness to be a member of my thesis
committee; your time and effort are greatly appreciated. Thank you to Ward
Dobbs and Zack Cicone for your excellent mentorship and guidance through-
out this process.
References
1. Buchheit M. Monitoring training status with HR measures: do
all roads lead to Rome? Front Physiol. 2014;5:73. PubMed ID:
24578692 doi:10.3389/fphys.2014.00073
2. Task Force of the European Society of Cardiology the North Ameri-
can Society of Pacing Electrophysiology. Heart rate variability:
standards of measurement, physiological interpretation, and clinical
use. Task force of the European society of cardiology and the North
American society of pacing and electrophysiology. Circulation.
1996;93(5):1043–1065. doi:10.1161/01.CIR.93.5.1043
3. Esco MR, Flatt AA. Ultra-short-term heart rate variability indexes at
rest and post-exercise in athletes: evaluating the agreement with
accepted recommendations. J Sports Sci Med. 2014;13(3):535–541.
PubMed ID: 25177179
4. Atlaoui D, Pichot V, Lacoste L, Barale F, Lacour J, Chatard J. Heart
rate variability, training variation and performance in elite swimmers.
Int J Sports Med. 2007;28(5):394–400. PubMed ID: 17111320
doi:10.1055/s-2006-924490
5. Garet M, Tournaire N, Roche F, et al. Individual interdependence
between nocturnal ANS activity and performance in swimmers. Med
Sci Sports Exerc. 2004;36(12):2112–2118. PubMed ID: 15570148
doi:10.1249/01.MSS.0000147588.28955.48
(Ahead of Print)
6Sherman et al
6. Iellamo F, Legramante JM, Pigozzi F, et al. Conversion from vagal to
sympathetic predominance with strenuous training in high-performance
world class athletes. Circulation. 2002;105(23):2719–2724. PubMed ID:
12057984 doi:10.1161/01.CIR.0000018124.01299.AE
7. Bellenger CR, Fuller JT, Thomson RL, Davison K, Robertson EY,
Buckley JD. Monitoring athletic training status through autonomic
heart rate regulation: a systematic review and meta-analysis. Sports
Med. 2016;46(10):1461–1486. PubMed ID: 26888648 doi:10.1007/
s40279-016-0484-2
8. Hagerman FC. Applied physiology of rowing. Sports Med. 1984;
1(4):303–326. PubMed ID: 6390606 doi:10.2165/00007256-
198401040-00005
9. Mäestu J, Jürimäe J, Jürimäe T. Monitoring of performance and training
in rowing. Sports Med. 2005;35(7):597–617. PubMed ID: 16026173
10. Flatt AA, Hornikel B, Esco MR. Heart rate variability and psycho-
metric responses to overload and tapering in collegiate sprint-
swimmers. J Sci Med Sport. 2017;20(6):606–610. PubMed ID:
27890479 doi:10.1016/j.jsams.2016.10.017
11. Plews DJ, Laursen PB, Stanley J, Kilding AE, Buchheit M.
Training adaptation and heart rate variability in elite endurance
athletes: opening the door to effective monitoring. Sports Med.
2013;43(9):773–781. PubMed ID: 23852425 doi:10.1007/s40279-
013-0071-8
12. Penttila J, Helminen A, Jartti T, et al. Time domain, geometrical and
frequency domain analysis of cardiac vagal outflow: effects of various
respiratory patterns. Clin Physiol. 2001;21(3):365–376. PubMed ID:
11380537 doi:10.1046/j.1365-2281.2001.00337.x
13. Flatt AA, Esco MR. Heart rate variability stabilization in athletes:
towards more convenient data acquisition. Clin Physiol Funct Imag-
ing. 2016;36(5):331–336. PubMed ID: 25754514 doi:10.1111/cpf.
12233
14. Aubert AE, Seps B, Beckers F. Heart rate variability in athletes.
Sports Med. 2003;33(12):889–919. PubMed ID: 12974657 doi:10.
2165/00007256-200333120-00003
15. Pereira LA, Flatt AA, Ramirez-Campillo R, Loturco I, Nakamura FY.
Assessing shortened field-based heart-rate-variability-data acquisition
in team-sport athletes. Int J Sports Physiol Perform. 2016;11(2):
154–158. PubMed ID: 26115088 doi:10.1123/ijspp.2015-0038
16. Schneider C, Hanakam F, Wiewelhove T, et al. Heart rate monitoring
in team sports—a conceptual framework for contextualizing heart rate
measures for training and recovery prescription. Front Physiol.
2018;9:639. PubMed ID: 29904351 doi:10.3389/fphys.2018.00639
17. Fatisson J, Oswald V, Lalonde F. Influence diagram of physiological
and environmental factors affecting heart rate variability: an extended
literature overview. Heart Int. 2016;11(1):32–40. doi:10.5301/
heartint.5000232
18. Janse de Jonge XA. Effects of the menstrual cycle on exercise
performance. Sports Med. 2003;33(11):833–851. PubMed ID:
12959622 doi:10.2165/00007256-200333110-00004
19. Plews DJ, Scott B, Altini M, Wood M, Kilding AE, Laursen PB.
Comparison of heart rate variability recording with smart phone
photoplethysmographic, polar H7 chest strap and electrocardiogram
methods. Int J Sports Physiol Perform. 2017;12(10):1324–1328.
PubMed ID: 28290720 doi:10.1123/ijspp.2016-0668
20. Esco MR, Flatt AA, Nakamura FY. Agreement between a smartphone
pulse sensor application and electrocardiography for determining
lnRMSSD. J Strength Cond Res. 2017;31(2):380–385. PubMed
ID: 28125545 doi:10.1519/JSC.0000000000001519
21. Flatt AA, Esco MR, Allen JR, et al. Heart rate variability and training
load among national collegiate athletic association division 1 college
football players throughout spring camp. J Strength Cond Res. 2018;
32(11):3127–3134. doi:10.1519/JSC.0000000000002241
22. Nakamura FY, Pereira LA, Cal Abad CC, et al. Adequacy of the
ultra-short-term HRV to assess adaptive processes in youth female
basketball players. J Hum Kinet. 2017;56(1):73–80. PubMed ID:
28469745 doi:10.1515/hukin-2017-0024
23. Mikuli´c P, Smoljanovi ´c T, Bojani´c I, Hannafin JA, Matkovi´c BR.
Relationship between 2000-m rowing ergometer performance times
and World Rowing Championships rankings in elite-standard rowers.
J Sports Sci. 2009;27(9):907–913. doi:10.1080/02640410902911950
24. Nakamura FY, Pereira LA, Esco MR, et al. Intraday and interday
reliability of ultra-short-term heart rate variability in rugby union
players. J Strength Cond Res. 2017;31(2):548–551. PubMed ID:
27243917 doi:10.1519/JSC.0000000000001514
25. Plews D, Laursen P, Le Meur Y, Hausswirth C, Kilding A, Buchheit
M. Monitoring training with heart rate-variability: how much com-
pliance is needed for valid assessment? Int J Sports Physiol Perform.
2014;9(5):783–790. PubMed ID: 24334285 doi:10.1123/ijspp.2013-
0455
26. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive
statistics for studies in sports medicine and exercise science. Med Sci
Sports Exerc. 2009;41(1):3–13. PubMed ID: 19092709 doi:10.1249/
MSS.0b013e31818cb278
27. Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet. 1986;
1(8476):307–310. PubMed ID: 2868172 doi:10.1016/S0140-
6736(86)90837-8
28. Flatt AA, Esco MR. Smartphone-derived heart-rate variability and
training load in a women’s soccer team. Int J Sports Physiol Perform.
2015;10(8):994–1000. PubMed ID: 25756657 doi:10.1123/ijspp.
2014-0556
29. Bonnemeier H, Wiegand Uwe KH, Brandes A, et al. Circadian
profile of cardiac autonomic nervous modulation in healthy subjects.
J Cardiovasc Electrophysiol. 2003;14(8):791–799. PubMed ID:
12890036 doi:10.1046/j.1540-8167.2003.03078.x
30. Molgaard H, Sorensen KE, Bjerregaard P. Circadian variation and
influence of risk factors on heart rate variability in healthy subjects.
Am J Cardiol. 1991;68(8):777–784. PubMed ID: 1892086
(Ahead of Print)
HRV Recording Time and Performance in Female Rowers 7
