The practical application of heart rate variability - monitoring training adaptation in world class athletes
Few will disagree that it is the preparatory exercise training completed that makes the most substantial impact on individual exercise performance. It is not surprising therefore that we often see elite athletes that chronically live between the borders of positive adaptation and maladaptation (overtraining/non-functional overreaching) to training, as they attempt to reach the greatest fitness level possible. The adaptive response to any number of training stimuli however, are individual, with genetic factors likely being a substantial determinant of how an athlete might respond. The ability to effectively track these individual responses (positive or negative) using quantitative physiological measures would be advantageous for sports practitioners and coaches alike. Heart rate variability (HRV) provides an estimate of a person’s cardiac autonomic activity, and has shown promise as a potential tool to monitor individual adaptation to endurance training. However, effective methods for assessment are yet to be established. Therefore, the overarching aim of this doctoral thesis was to establish methods by which vagally-derived indices of HRV can be practically applied to monitor and assess an elite endurance athlete’s adaptation to training in an every-day setting. In order to answer this question effectively, this thesis is made up of one case comparison, two methodological studies, one current opinion and an observational study.
In the first study of the thesis, a case comparison, the daily HRV and training (23 h ± 2 h per week) were monitored over a 77-day period in two elite triathletes (one male: 22 yr, V ̇O2max 72.5 ml.kg.min-1; one female: 20 yr, V ̇O2max 68.2 ml.kg.min-1). During this period, one athlete performed poorly in a key triathlon event and was diagnosed as non-functionally over-reached (NFOR). The 7-day rolling average of the log-transformed square root of the mean sum of the squared differences between R-R intervals (Ln rMSSD) were compared to the individual smallest worthwhile change (SWC). Ln rMSSD values declined towards the day of the triathlon event (slope = -0.17 ms/wk; r2 = -0.88) in the NFOR athlete, and remained stable in the control athlete (slope = 0.01 ms/wk; r2 = 0.12). Furthermore, in the NFOR athlete, the HRV coefficient of variation (CV of Ln rMSSD 7-day rolling average) revealed large linear reductions towards NFOR (i.e., linear regression of HRV variables vs. day number towards NFOR: -0.65 %/wk and r2 = -0.48), while these variables remained stable for the control athlete (slope = 0.04 %/wk). These data suggest that trends in both absolute HRV values and day-to-day variations along with the individual SWC may be useful measurements indicative of the progression towards mal-adaptation or non-functional over-reaching.
Considering the findings of the case comparison, showing that weekly-averaged Ln rMSSD values provided superior representation of maladaptation compared with values taken on a single day, the aim of the second study of the thesis was to compare relationships between performance, positive adaptation and HRV measured on an isolated day or with values averaged over the week. The relative change in estimated maximum aerobic speed (MAS) and 10-km running performance was correlated to the relative change in Ln rMSSD on an isolated day (Ln rMSSDday) or when averaged over 1 week (Ln rMSSDweek) in 10 runners who responded to a 9-week training intervention. A trivial correlation was observed for MAS vs. Ln rMSSDday (r = -0.06 (-0.59; 0.51)), while a very-large correlation was shown between MAS and Ln rMSSDweek (r = 0.72 (0.28; 0.91)). Similarly, changes in 10-km running performance revealed a small correlation with Ln rMSSDday (r = -0.17 (-0.66; 0.42)), versus a very-large correlation for Ln rMSSDweek (r = -0.76 (-0.92; -0.36)). It was concluded that the averaging of HRV values over a 1-week period appeared to be a superior method for monitoring positive adaption to training compared with assessing its value on a single isolated day.
The third study of the thesis was based on the findings from studies 1 and 2, and its primary aim was to establish the minimum number of days that Ln rMSSD data should be averaged before equivalent outcome results were attained. Standardised changes in Ln rMSSD between different phases of training (normal training, functional overreaching, overall training and taper) and the correlation coefficients were compared when averaging Ln rMSSD from 1 to 7 days, randomly selected within the week. Standardised Ln rMSSD changes (90% confidence intervals, CI) from baseline to overload (functional overreaching) were 0.20 (-0.08; 0.47); 0.33 (0.07; 0.59); 0.49 (0.17; 0.82); 0.48 (0.20; 0.76); 0.47 (0.21, 0.73); 0.45 (0.19; 0.71) and 0.43 (0.19; 0.72) using from 1 to 7 days, respectively. Correlations (90% CI) over the same time sequence and training phase were: -0.02 (-0.21; 0.25); -0.07 (-0.16; 0.3); -0.17 (-0.16; 0.3); -0.25 (-0.45; -0.02); -0.26 (-0.46; -0.03); -0.28 (-0.48; -0.5) and -0.25 (-0.45; -0.2) from 1 to 7 days, respectively. There were almost perfect quadratic relationships between standardised changes/r values vs. the number of days Ln rMSSD was averaged (r2 = 0.92 and 0.97, respectively), indicating a plateau in the increase in the magnitude of the standardised changes/r values after 3 and 4 days, respectively, in trained triathletes. It was concluded that practitioners using HRV to monitor training adaptation in trained athletes should use a minimum of 3 (randomly selected) valid data points per week.
While assessing HRV in a number of elite athletes, it became clear to me that a shift in current opinion on various issues was required. Accordingly, the fourth study in this thesis, a current opinion review, outlines the changes in HRV in response to training loads and the likely positive and negative adaptations shown, along with some limitations to these reported findings. Solutions are offered to some of the methodological issues associated with using HRV as a day-to-day monitoring tool, including the use of appropriate averaging techniques, and the Ln rMSSD to R-R interval ratio to overcome the issue of HRV saturation in elite athletes (i.e. reductions in HRV despite decreases in resting heart rate). Finally, this work offers examples in Olympic and World Champion athletes, showing how these indices can be practically applied to assess training status and readiness to perform in the period leading up to a pinnacle event. The paper reveals how longitudinal HRV monitoring in elites is required to understand their unique individual HRV fingerprint. For the first time, it is demonstrated how increases and decreases in HRV relate to changes in fitness and freshness, respectively, in elite athletes.
In the final study of the thesis, the relationship between HRV and training intensity distribution in elite rowers (4 female, 5 male) were examined during a 26-week build-up to the 2012 Olympic Games. The weekly-averaged Ln rMSSD were reported, and compared to changes in total training time (TTT) and training time below the first lactate threshold (<LT1); above the second lactate threshold (LT2), and between LT1 and LT2 (LT1-LT2). After substantial increases in training time in a particular training zone/load variable (average effect size = 1.47, 90% confidence limits (1.35; 1.59)), standardized changes in Ln rMSSD were +0.13 (trivial; unclear) for TTT, +0.20 (small; 51% chance of greater values) for time <LT1, -0.02 (trivial and unclear) for time LT1-LT2, and -0.20 (small; 53% chance of lower values) for time >LT2. Correlations for Ln rMSSD were small vs. TTT (r = 0.37 (0.28; 0.45)), moderate vs. time <LT1 (r =0.43 (0.32; 0.53)), trivial vs. LT1-LT2 (r = 0.01 (-0.16; 0.17)) and small vs. >LT2 (r = -0.22 (-0.27; -0.17)). These data suggest that training phases with increased time spent at high intensity suppress cardiac parasympathetic activity, whilst low-intensity training preserves and increases it. Practically, ~5% increase in high-intensity training should be accompanied by ~6% increase in low-intensity training, so that autonomic balance is preserved.
Collectively, the studies in this thesis demonstrate that vagally-derived indices of HRV can be used as an effective tool to individually monitor endurance training in elite athletes. The new findings included: 1) that optimal monitoring should be carried out using Ln rMSSD values averaged over a minimum of 3 days per week (or 1 micro-cycle) alongside the individual SWC, 2) that the Ln rMSSD to R-R interval ratio should also be measured when considering changes due to training as a result of HRV saturation present in elite athletes, and 3) that when monitoring training using HRV, changes should be considered in light of the training phase being completed; that is, that increases in HRV falling above the SWC during periods of overload are likely reflective of positive adaptation, and decreases in HRV below the SWC with reduced training loads (e.g. taper) are likely signs of increasing freshness and readiness to perform. HRV values falling below the SWC, coupled with substantial increases in the Ln rMSSD to R-R interval ratio during periods of high training loads may be indicative of maladaptation.