I recently had the pleasure of discussing HRV in football and rugby on the Rugby Renegade Podcast. Soundcloud and iTunes links below.
I recently had the pleasure of discussing HRV in football and rugby on the Rugby Renegade Podcast. Soundcloud and iTunes links below.
I’ve recently had the pleasure of peer-reviewing a few very well-written and carried out studies investigating duration requirements for stabilization preceding HRV recordings by different research groups. I look forward to seeing the published versions as the quality of the papers was very high.
In reviewing these papers it prompted me to reconsider what we all have been using as the criterion period. My colleagues and I have published 5 papers using a 5-min R-R sample preceded by a 5-min ‘stabilization’ period (10 min total duration) as the criterion (as has other groups), which is in line with traditional procedures. But I think we failed to address an important limitation of these procedures…
The issue is that the ‘traditional procedures’ were not devised for the purposes of establishing LnRMSSD specifically (rather, they needed to accommodate spectral analysis), nor were they devised for reflecting fatigue and adaptation to training programs. Therefore, for these specific purposes, it can be argued that the traditional procedures may not be as relevant, or at the very least, calls into question whether the 5-10 min period following the 0-5 min stabilization is in fact a criterion within this context.
Some things to consider:
To be clear, I still think that research evaluating stabilization requirements and comparing to the ‘criterion’ is absolutely meaningful and an important starting point. This was not intended to be critical, but rather to open discussion on future research directions.
Here’s a brief overview of our latest study capturing daily HRV and wellness ratings throughout overload and tapering in collegiate sprint swimmers preceding a championships competition.
The majority of research in the area has primarily focused on endurance athletes. It’s been a goal of mine for a while now to examine HRV responses in athletes participating in anaerobic events such as short-distance swimming.
The athletes completed wellness questionnaires and recorded HRV daily via smartphone and validated pulse-wave finger sensor (seated position) after waking. The observation period lasted 6-weeks which included 1 week of baseline, 2 weeks of overload and a progressive 3-week taper. The overload was characterized by a substantial increase in training intensity while overall volume varied by up to only 20%. Of the group, 2 athletes went on to compete in the 2016 Olympic summer games.
The purpose of this study was to evaluate cardiac-parasympathetic and psychometric responses to competition preparation in collegiate sprint-swimmers. Additionally, we aimed to determine the relationship between average vagal activity and its daily fluctuation during each training phase.
Ten Division-1 collegiate sprint-swimmers performed heart rate variability recordings (i.e., log transformed root mean square of successive RR intervals, lnRMSSD) and completed a brief wellness questionnaire with a smartphone application daily after waking. Mean values for psychometrics and lnRMSSD (lnRMSSDmean) as well as the coefficient of variation (lnRMSSDcv) were calculated from 1 week of baseline (BL) followed by 2 weeks of overload (OL) and 2 weeks of tapering (TP) leading up to a championship competition.
Competition preparation resulted in improved race times (p<0.01). Moderate decreases in lnRMSSDmean, and Large to Very Large increases in lnRMSSDcv, perceived fatigue and soreness were observed during the OL and returned to BL levels or peaked during TP (p<0.05). Inverse correlations between lnRMSSDmean and lnRMSSDcv were Very Large at BL and OL (p<0.05) but only Moderate at TP (p>0.05).
OL training is associated with a reduction and greater daily fluctuation in vagal activity compared with BL, concurrent with decrements in perceived fatigue and muscle soreness. These effects are reversed during TP where these values returned to baseline or peaked leading into successful competition. The strong inverse relationship between average vagal activity and its daily fluctuation weakened during TP.
While group responses are certainly meaningful, the individual responses provide more meaningful information to practitioners. The figure below shows the individual trends from 3 athletes that exemplify 3 common training responses I’ve observed in a variety of athletes.
Subject B (middle) has the smallest CV at baseline and subsequently handles the overload very well, with minimal reductions in lnRMSSD. This indicates that Subject B is in great shape and could probably handle greater loads.
Subject C (bottom) displays what I would consider to be a very expected response to overloading. There is a considerable increase in daily lnRMSSD fluctuation (i.e., increased CV) and progressive but small decrease in the trend. I interpret this type of response to indicate that the loads are sufficient enough to provoke the fatigue/recovery process but not so high that HRV becomes suppressed. This is possibly indicative of a load/dose of training that is high but within the overall recovery capacity of the athlete.
Subject A (top) has the highest CV of the group at baseline and subsequently responds the least favorably to the overload. lnRMSSD pretty much crashes almost immediately and remains suppressed for several days (boxed data points). The coach pulled back on subject A due to high fatigue, reduced performance, decrements in pulse-rate recovery between sets, etc. The trend immediately improves until about 1-week out from competition at which point loads again were further reduced. Ultimately, this athlete improved upon previous best times at competition from that year, suggesting that the interventions were effective.
The main take-home would be that the typical response to intensified training includes a reduction and greater daily fluctuation in HRV, along with decrements in wellness scores. Athletes demonstrating different responses (i.e., minimal change in HRV trend or conversely chronic suppression of HRV) may be coping better or worse than expected. Coaches should then investigate and address factors contributing to the poor response.
When implementing HRV monitoring with a new team, the coach will be quick to point out the inter-individual variability in the athletes’ trends. Some athletes are showing high scores and some are low. Some are showing considerable daily fluctuation while others show very consistent numbers. Or, some show substantial fluctuation during this period but minimal fluctuation during that period. This can be confusing and difficult to interpret, but with some context, the trends (and changes therein) can usually be explained.
Greater fitness levels are associated with higher resting HRV and faster parasympathetic reactivation following exercise. This likely contributes to the smaller coefficient of variation (CV) we (and others) have observed in athletes with higher VO2max and intermittent running performance. So if we were to categorize athletes of the same sport based on competitive level (i.e., training status), we should see group differences between their average lnRMSSD and CV. What makes our approach different from previous work is the longer observation period (1 month), the use of a finger sensor (PPG) and smartphone application using ultra-short HRV recordings for daily data acquisition and inclusion of the CV in the analysis. This was presented at the NSCA National Conference in New Orleans this July. Full manuscript in production soon.
THE EFFECT OF TRAINING STATUS ON HEART RATE VARIABILITY IN DIVISION-1 COLLEGIATE SWIMMERS
Andrew A. Flatt, Bjoern Hornikel, Michael R. Esco
University of Alabama, Tuscaloosa, AL
Resting heart rate variability (HRV) fluctuates on a daily basis in response to physical and psychological stressors and may provide useful information pertaining to fatigue and adaptation. However, there is limited research comparing HRV profiles between athletes of the same sport who differ by training status. PURPOSE: The purpose of this study was to compare resting heart rate (RHR) parameters between national and conference level Division-1 Collegiate swimmers and to determine if any differences were related to psychometric indices. METHODS: Twenty-four subjects were categorized as national (NAT, n = 12, 4 female) or conference level competitors (CONF, n=12, 5 female). Over 4 weeks, daily HRV was measured in the seated position by the subjects after waking and elimination with a validated smartphone application and pulse-wave finger sensor (app) utilizing a 55-second recording period. Subjects then completed a questionnaire on the app where they rated perceived levels of sleep quality, muscle soreness, mood, stress and fatigue on a 9-point scale. The HR parameters evaluated by the app include RHR and the log-transformed root-mean square of successive RR interval differences multiplied by 20 (lnRMSSD). The 4-week mean for RHR (RHRm) and lnRMSSD (lnRMSSDm) in addition to the coefficient of variation (CV) for RHR (RHRcv) and lnRMSSD (lnRMSSDcv) were determined for comparison. In addition, psychometric parameters were also averaged between groups and compared. Independent t-tests and effect sizes ± 90% confidence limits (ES± 90% CL) were used to compare the HR and psychometric parameters. RESULTS: NAT was moderately taller (184.9 ± 10.0 vs. 175.5 ± 12.5 cm; p = 0.06, ES ± 90% CL = 0.83 ± 0.70) and heavier (80.4 ± 9.7 vs. 75.2 ± 11.9 kg; p = 0.26, ES ± 90% CL = 0.48 ± 0.67) than CONF, though not statistically significant. The results comparing HR and psychometrics are displayed in Table 1. lnRMSSDm and lnRMSSDcv was moderately higher and lower, respectively, in NAT compared to CONF (p<0.05). CONCLUSION: Higher training status is associated with moderately higher lnRMSSDm and lower lnRMSSDcv compared to those of lower training status. This was observed despite no significant difference in perceived stressors that may affect HR parameters. PRACTICAL APPLICATION: Training status appears to be a determinant of daily HRV and its fluctuation. This may be because higher level athletes are more fit and recover faster from training, resulting in a more stable HRV pattern. This information can be useful to practitioners when interpreting HRV trends in athletes. For example, an increase in HRV with reduced daily fluctuation may indicate improvements in an athletes training status. Alternatively, an athlete with high training status demonstrating reduced HRV and greater daily fluctuation may be showing signs of fatigue or loss of fitness depending on the context of the current training phase and program.
This figure shows a year of data from two athletes (Olympic level on top vs. Conference level on bottom) to provide a nice visual representation of their trend differences.
Here’s a brief overview of our latest study, in press with JSCR. We compared the ithlete finger sensor with ECG in supine, seated and standing positions. We are continuing our testing with other popular smartphone HRV apps in the near future. Thanks to the Summer 2015 Alabama S&C interns for making up a large portion of the participants in this study.
The purpose of this study was to determine the agreement between a smartphone pulse finger sensor (SPFS) and electrocardiography (ECG) for determining ultra-short-term heart rate variability (HRV) in three different positions. Thirty college-aged men (n = 15) and women (n = 15) volunteered to participate in this study. Sixty second heart rate measures were simultaneously taken with the SPFS and ECG in supine, seated and standing positions. lnRMSSD was calculated from the SPFS and ECG. The lnRMSSD values were 81.5 ± 11.7 via ECG and 81.6 ± 11.3 via SPFS (p = 0.63, Cohen’s d = 0.01) in the supine position, 76.5 ± 8.2 via ECG and 77.5 ± 8.2 via SPFS (p = 0.007, Cohen’s d = 0.11) in the seated position, and 66.5 ± 9.2 via ECG and 67.8 ± 9.1 via SPFS (p < 0.001, Cohen’s d = 0.15) in the standing positions. The SPFS showed a possibly strong correlation to the ECG in all three positions (r values from 0.98 to 0.99). In addition, the limits of agreement (CE ± 1.98 SD) were -0.13 ± 2.83 for the supine values, -0.94± 3.47 for the seated values, and -1.37 ± 3.56 for the standing values. The results of the study suggest good agreement between the SPFS and ECG for measuring lnRMSSD in supine, seated, and standing positions. Though significant differences were noted between the two methods in the seated and standing positions, the effect sizes were trivial.
Full Text on Research Gate
Here’s a look at our latest study in collaboration with Fabio Nakamura and colleagues, now in press with JSCR (Abstract below). In this study, HRV was recorded as a team at the training facility, not immediately after waking. This is the approach that many coaches are interested in using given the issue with compliance when trying to get athletes to perform HRV measures on their own at home after waking. Controlled and supervised measures at the facility appear promising, at least in these high level athletes.
It’s important to understand that autonomic activity is constantly making adjustments to physical, chemical and perceived psychological stimuli. Thus, HRV is inherently not the most reliable metric. However, training status/fitness appear to have a strong affect on day to day variation in HRV. More fit athletes recover faster/tolerate training better and thus tend to show less deviation from baseline compared to less fit athletes, of which will experience much greater homeostatic disruption from training and greater day to day variation. I strongly believe that the amount of daily fluctuation (i.e., lnRMSSDcv) is a very useful indication of fitness, stress and training adaptation.
We currently have a paper in production looking at the effect of training status on HRV. In the mean time, compare the trends below of an Olympic level and a conference level athlete, both short-distance swimmers (similar age and physical characteristics) across 4 consecutive weeks of training.
The aim of this study was to examine the intra-day and inter-day reliability of ultra-short-term vagal-related heart rate variability (HRV) in elite rugby union players. Forty players from the Brazilian National Rugby Team volunteered to participate in this study. The natural log of the root mean square of successive RR interval differences (lnRMSSD) assessments were performed on four different days. HRV was assessed twice (intra-day reliability) on the first day and once per day on the following three days (inter-day reliability). The RR interval recordings were obtained from 2-min recordings using a portable heart rate monitor. The relative reliability of intra- and inter-day lnRMSSD measures were analyzed using the intraclass correlation coefficient (ICC). The typical error of measurement (absolute reliability) of intra- and inter-day lnRMSSD assessments were analyzed using the coefficient of variation (CV). Both intra-day (ICC = 0.96; CV = 3.99%) and inter-day (ICC = 0.90; CV = 7.65%) measures were highly reliable. The ultra-short-term lnRMSSD is a consistent measure for evaluating elite rugby union players, in both intra- and inter-day settings. This study provides further validity to using this shortened method in practical field conditions with highly trained team sports athletes.
Full text on Research Gate
Here’s a brief look at a new paper of ours in press with JSCR. This is a very small study that we submitted as “Research Note” that looked at changes in HRV (via finger pulse sensor) and training load (via Polar Team2) across preseason training in D-1 women’s soccer players.
The purpose of this study was to track changes in training load (TL) and recovery status indicators throughout a 2-week preseason and to interpret the meaning of these changes on an individual basis among 8 Division-1 female soccer players. Weekly averages for heart ratevariability (lnRMSSD), TL and psychometrics were compared with effect sizes (ES) and magnitude based inferences. Relationships were determined with Pearson correlations. Group analysis showed a very likely moderate decrease for total training load (TTL) (TTL week 1 = 1203 ± 198, TTL week 2 = 977 ± 288; proportion = 1/2/97, ES = -0.93) and a likely small increase in lnRMSSD (week 1 = 74.2 ± 11.1, week 2 = 78.1 ± 10.5; proportion = 81/14/5, ES = 0.35). Fatigue demonstrated a very likely small improvement (week 1 = 5.03 ± 1.09, week 2 = 5.51 ± 1.00; proportion = 95/4/1; ES = 0.45) while the other psychometrics did not substantially change. A very large correlation was found between changes in TL and lnRMSSD (r = -0.85) while large correlations were found between lnRMSSD and perceived fatigue (r = 0.56) and soreness (r = 0.54). Individual analysis suggests that 2 subjects may benefit from decreased TL, 2 subjects may benefit from increased TL and 4 subjects may require no intervention based on their psychometric and lnRMSSD responses to the TL. Individual weekly changes in lnRMSSD varied among subjects and related strongly with individual changes in TL. Training intervention based on lnRMSSD and wellness responses may be useful for preventing the accumulation of fatigue in female soccer players.
Full Text on Research Gate
This study aimed to compare the weekly natural log of the root-mean-square difference of successive normal inter-beat (RR) intervals (ln RMSSDWeekly) and its coefficient of variation (ln RMSSDCV) in response to 5 weeks of preseason training in professional male futsal players. A secondary aim was to assess the relationship between ln RMSSDWeekly and ln RMSSDCV. The ln RMSSD is a measure of cardiac–vagal activity, and ln RMSSDCV represents the perturbations of cardiac autonomic homeostasis, which may be useful for assessing how athletes are coping with training. Ten futsal players had their resting ln RMSSD recorded prior to the first daily training session on four out of approximately five regular training days·week−1. Session rating of perceived exertion (sRPE) was quantified for all training sessions. Despite weekly sRPE varying between 3455 ± 300 and 5243 ± 463 arbitrary units (a.u.), the group changes in ln RMSSDWeekly were rated as unclear (using magnitude-based inference), although large inter-individual variability in ln RMSSD responses was observed. The ln RMSSDCV in weeks 4 and 5 were likely lower than the previous weeks. A large and significant negative correlation (r = −0.53; CI 90%: −0.36; −0.67) was found between ln RMSSD and ln RMSSDCV. Therefore, monitoring individual ln RMSSD responses is suggested since large inter-individual variations may exist in response to futsal training. In addition, higher values of ln RMSSD are associated with lower oscillations of cardiac autonomic activity.
Full Text on Research Gate
I recently had a chance to sit down and discuss all things HRV monitoring with James Darley of the Historic Performance Podcast. There’s also a number of great interviews in the podcast archives worth checking out.
Here’s a quick look at our latest study published ahead of print in the Journal of Sports Medicine and Physical Fitness. The full text is available here. Below is the abstract and some brief comments about the findings.
BACKGROUND: Heart rate variability (HRV) is an objective physiological marker that may be useful for monitoring training status in athletes. However, research aiming to interpret daily HRV changes in female athletes is limited. The objectives of this study were (1) to assess daily HRV (i.e., log-transformed root mean square of successive R-R interval differences, lnRMSSD) trends both as a team and intra-individually in response to varying training load (TL) and (2) to determine relationships between lnRMSSD fluctuation (coefficient of variation, lnRMSSDcv) and psychometric and fitness parameters in collegiate female soccer players (n=10).
METHODS: Ultra-short, Smartphone-derived lnRMSSD and psychometrics were evaluated daily throughout 2 consecutive weeks of high and low TL. After the training period, fitness parameters were assessed.
RESULTS: When compared to baseline, reductions in lnRMSSD ranged from unclear to very likely moderate during the high TL week (effect size ± 90% confidence limits [ES ± 90% CL] = -0.21 ± 0.74 to -0.64 ± 0.78, respectively) while lnRMSSD reductions were unclear during the low TL week (ES ± 90% CL = -0.03 ± 0.73 to -0.35 ± 0.75, respectively). A large difference in TL between weeks was observed (ES ± 90% CL = 1.37 ± 0.80). Higher lnRMSSDcv was associated with greater perceived fatigue and lower fitness (r [upper and lower 90% CL] = -0.55 [-0.84, -0.003] large, -0.65 [-0.89, -0.15] large).
CONCLUSIONS: Athletes with lower fitness or higher perceived fatigue demonstrated greater reductions in lnRMSSD throughout training. This information can be useful when interpreting individual lnRMSSD responses throughout training for managing player fatigue.
The idea of evaluating relationships between the coefficient of variation of lnRMSSD (lnRMSSDcv) with fitness parameters was inspired by a 2010 paper by Martin Buchheit et al. In that study, greater lnRMSSDcv derived from post-submaximal exercise recordings negatively correlated with maximum aerobic speed in youth soccer players. We had similar findings in our current paper where we observed large negative relationships between lnRMSSDcv (derived from waking, ultra-short smartphone recordings) and VO2max and Yo-Yo IRT-1.
Another objective of this study was to focus on individual HRV responses in addition to group responses (see figure below). An interesting observation we made was that greater lnRMSSDcv was also associated with higher perceived fatigue. This finding is in contrast to a recent case comparison study by Plews et al. that found a decreased lnRMSSDcv to be associated with non-functional overreaching in an elite triathlete. However, this can possibly be explained by the severity of fatigue. For example, the decreased lnRMSSDcv observed in the triathlete was accompanied with a chronically suppressed lnRMSSDmean. Thus, lnRMSSD decreased and did not periodically return to baseline.
In our current study, large decreases in lnRMSSD typically returned to baseline after 24-72 hours. Thus, loads were not so high that the athletes were unable to return to baseline. Therefore, it is possible that there may be a progression in one’s HRV trend leading from moderately fatigued to severely fatigued that is characterized first by a greater lnRMSSDcv (reflecting fatigue and recovery process) followed by chronic suppression of lnRMSSD with no rebounding to baseline (reduced lnRMSSDmean and reduced lnRMSSDcv). More on this to come.