Heart rate variability, mood and performance: a pilot study on the interrelation of these variables in amateur road cyclists

Sample

Five recreational road cyclists volunteered to participate in this study. All were males with an average of 31.6 ± 3.4 years, had at least 6 years of experience riding and completed at least two training sessions per week for the duration of the study. Each cyclist was fully informed of potential risks and benefits associated with participation. Written informed consent was obtained from each participant. The study was conducted according to the Local Ethics Commission for Human Experimentation of the Autonomous University of Barcelona (protocol code CEEAH-5745).

Instruments

Procedure

A longitudinal case study design was used to compare HRV and performance responses of the volunteers during a 6-week period. All eligible participants attended a first face-to-face interview individually in which demographic and fitness indices were collected. The protocol of the study was explained. After the first interview, two sets of data were recorded, according to the individual training planning of each cyclist:

• a)

  Training data: during the 6-week period, each participant trained individually, following their own training plan, which consisted of riding road bikes at least twice a week. The study didn’t interfere with their training routine. For each training session, power data was recorded and was used to calculate TSS and IF using a power meter. RPE was also recorded. Data was saved and processed using the app TrainingPeaks.

• b)

  Morning data: After days of training, at home on their own, upon waking on the morning, in a supine position in bed, and prior to any movement or ingest, each participant answered a 10-point mood scale. Immediately afterwards, maintaining the supine position in bed, each participant was required to complete a 3 min HRV test. Data was collected and analysed with the FitLab® system (HealthSportLab.com, Barcelona, Spain).

Participants were asked to complete the maximum number of training sessions per week possible, followed by recording morning data the day after. A total of 123 morning registers (mood and 3 min HRV test) and a total of 66 training registers were recorded. A total of 57 HRV morning recordings were not preceded by a road training the day before. During the 6-week study period, each participant completed an average of 25 morning registers (4 per week) and 12 training registers (2 per week).

HRV data analysis

HRV analysis was performed for RR intervals in 3 min periods, as carried out in other field studies where it was required to interfere as little as possible in the training routines of athletes (Martín-Guillaumes et al., 2018). Ultra-short-term HRV analysis (30 s or 60 s) was recommended at rest condition (Wu et al., 2020), with a minimum of 120 s needed to record low frequency components (Munoz et al., 2015).

HRV data was obtained from the FitLab system. All RR intervals obtained were filtered and records exceeding an error rate of 11% were not considered for analysis. The mean of signal error for all recorded RR was 0.91%. Matlab’s scripts (MathWork, Portola Valley, CA, USA) were used for error correction, as developed and validated in previous publications (García-González et al., 2015; Parrado et al., 2010). The error correction of RR series was applied as defined by García-González et al. (2015). HRV analyses were performed following the recommendations of the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (TFESC, 1996). HRV parameters were calculated as defined by Escorihuela et al. (2020): for time domain analyses, mean of RR intervals (mean RR) and the standard deviation of all RR intervals (SDNN) were calculated, and for frequency domain analysis the power in the low frequency (LF) band (0.04–0.15 Hz) and the high frequency (HF) band (0.15–0.40 Hz) were used. Additional calculations included the normalized LF and HF values (LFnu and HFnu, respectively), and LF/HF parameters.

Statistical analysis

Analyses were carried out with all recordings for the five cyclists. Descriptive data are presented as mean ± SD. Pearson correlation analyses were performed in order to show bivariate relationships between HRV, mood, RPE, and external training load parameters. Multiple regression analysis was applied for explaining mood and HRV as a function of training parameters from the previous day, as well as mood as a function of resting HRV parameters in the same morning session. The regression method used was stepwise. G*Power (v3.1; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) was used to analyse statistical power and effect size for regression analysis (Faul et al., 2009). Other statistical analyses were performed using IBM SPSS Statistics Package for Mac OS (version 28.0; SPSS Inc., Chicago, IL, USA). Statistical significance was set at p < 0.05. All data are presented as mean ± SD unless otherwise stated.

Description of anthropometric data and training indices

The cycling group consisted of five male recreational road cyclists. Anthropometric characteristics and training indices are shown in Table 1.

Description of the main parameters analysed in the study

Table 2 shows means and standard deviations for the parameters analysed in the two situations under study, Training and Morning.

Correlation coefficients and regression analyses

The following tables show the correlation between variables registered in the morning and those during training completed on the previous day.

Yesterday’s training correlates with this morning’s HRV

The current study indicated a relationship between morning HRV and training from the previous day. Specifically, Table 3 showed an inverse relationship between IF and HFnu, and a positive relationship between IF and LF/HF. Figure 1 illustrates, for Participant 1, the joint evolution over a month of the transformed parameters HFnu and IF. It can be seen, for example, that both pieces of data have an upward trend between 15-Jul and 23-Jul, a downward trend between 31-Jul and 4-Aug, and an opposite trend from 7-Jul to 8-Jul. The remaining training parameters under study, including NP and TSS, followed the same pattern as IF (p < 0.01), but with a lower significance (p < 0.05). Since Hfnu is considered a modulation index of the ANS parasympathetic branch (Burr, 2007), the results suggested that a higher training load one day (higher IF) correlated negatively with the activation of the vagal nerve the following morning, in favour of the SNS. This is in line with other studies, showing that parasympathetic power, indicated by HF, is able to reflect the recovery status hours after training (Chen et al., 2011; Seiler, Haugen & Kuffel, 2007), and that increased exercise intensity and/or duration cause delayed recovery of nocturnal cardiac autonomic modulation (Myllymäki et al., 2012).

As mentioned below (see Future Research), it would be interesting to track HFnu at regular hours and days post-exercise to help determine whether training was well-tolerated by the athlete, and in particular, to identify at what point the ANS switched from a sympathetic to a parasympathetic dominance. This is because the speed at which ANS recovers after exercise seems to be determined by fitness levels, with more rapid recovery in highly trained than in trained subjects after high-intensity exercise (Seiler, Haugen & Kuffel, 2007). Also, an increase of the parasympathetic branch following exercise is suggested to indicate the readiness of the ANS to take on strain again (Kiviniemi et al., 2007; Mueck-Weymann, Janshoff & Mueck, 2004) and is associated with improved athletic performance (Buchheit et al., 2010; Garet et al., 2004).

Yesterday’s training correlates with this morning’s mood

Results showed an inverse relationship between IF and mood, so that the tougher one training session, the lower the mood of the following day (Table 4). This is contrary to the popular belief and most research on the topic, which argues that that exercise improves mood. However, such belief and research look at mood minutes, not days, after physical activity (Hansen, Stevens & Richard Coast, 2001; Lane & Lovejoy, 2001; Walters & Kearsley, 1993). Research looking at wellness scores the following day of exercise concluded that the greater the training load, the worse the wellness measures (fatigue, sleep quality, soreness and stress) (Buchheit et al., 2013). This goes in line with the results in Table 3. Other studies looking at mood over longer periods of time indicate that mood is affected by intense physical training in a dose-dependent manner, so that mood disturbances, as measured by the Profile of Mood States (POMS), increase in athletes during high-volume training sessions, in proportion to the training stimulus, and return to baseline during recovery (Gross et al., 2016; Purvis, Gonsalves & Deuster, 2010). At the same time, an impaired mood state and subjective complaints have mostly been described as sensitive and early markers of overtraining (Morgan et al., 1988; Urhausen & Kindermann, 2002), but not in all cases (Walters & Kearsley, 1993). While there is no clear way to define or diagnose overtraining, it is mostly characterised by ‘sports-specific’ decrease in performance, accompanied disturbances in mood state, including increase in fatigue, confusion, tension, depression, anxiety and lack of motivation and irritability, amongst others, persisting despite a period of recovery lasting several weeks or months (Meeusen et al., 2013; Stone et al., 1991; Walters & Kearsley, 1993).

In the present study it was found that the tougher one training session, the lower the mood of the following day, but it cannot be stated whether the change in mood was due to the intensity level of the training sessions nor from a state of overtraining. However, this is certainly a topic to be explored in future lines of research. Understanding short and long-term changes would be of interest to establish under which circumstances mood may be reliable and efficient to monitoring overtraining.

Morning mood as a function of HRV

Findings in the current study indicate a relationship between morning mood and HRV, consistent with the idea that emotions that humans experience are associated with varying degrees of physiological arousal (Levenson, 2003). It was found that RRmean, SDNN and HFnu, in conjunction, explain 49.2% of mood variability of the same day (Table 5). Table 4 showed that mood correlated positively with RRmean and HFnu, and inversely with LFnu, LF/HF and SDRR (p < 0.005). Consistent with the idea of Hfnu as an index of the PNS (Burr, 2007), the results suggested that the more relaxed or recovered the nervous system, the more positive the emotional state. The findings are in line with other studies indicating a positive correlation between mood and HRV (Kindermann, 1986; Kleiger, Stein & Bigger, 2005; Koval et al., 2013; Lee, Wood & Welsch, 2003; Sakuragi, Sugiyama & Takeuchi, 2002). Geisler et al. (2010) supports the idea that HRV is associated with subjective well-being, whereas reduced HRV can be used as a predictive factor of the development of negative moods after situations such as deprivation of exercise (Weinstein, Deuster & Kop, 2007). However, it is interesting to note that most studies that found a significant correlation between mood and HRV were exclusively considering positives moods, including vigour and situations that trigger laughing, instead of negative moods (Sakuragi, Sugiyama & Takeuchi, 2002; Yoshino & Matsuoka, 2011). This could be due to the fact that positive emotions have a strong but transient effect on the ANS, while sadder moods have moderate but sustained effect (Sakuragi, Sugiyama & Takeuchi, 2002). Therefore, it would be interesting to study the relationship between morning HRV and mood not only on the same day, but also in consecutive days.

RPE as an indicator of internal load

RPE, an internal load variable, correlated positively with external load variables such as NP, TSS and IF (p < 0.005). In this way, tougher workouts (higher power output) are reflected in higher RPE, supporting RPE as a significant measure of strain. This is consistent with other studies showing that session-RPE can be considered a valid, reliable and consistent indicator of global internal load (Coutts et al., 2003; Haddad et al., 2017; Impellizzeri et al., 2004; Impellizzeri, Rampinini & Marcora, 2005). This method was initially proposed by Foster et al. (1996) for monitoring internal training load in endurance athletes, but more recently is has also been proven to be applicable to other sports and physical activities with both men and women of different age and among different levels of experience (Haddad et al., 2017; Lupo, Ungureanu & Brustio, 2020; Wallace, Slattery & Coutts, 2009).

Limitations

The main limitation of the current study is that despite having a number of recordings of n = 123 for Morning data and n = 66 for Training data, it only corresponds to five cyclists. A larger number of participants would be required in order to minimize individual differences and to increase the percentage of variability explained by the models of correlations and regression. Moreover, the lack of repeated measures fails to detect changes in HR and mood across different training situations, and future studies could include a follow-up assessment to capture the effects of different situations on the athletes, as proposed by Olmedilla et al. (2018). Another limitation of the study is the use of a non-standardized scale to measure morning mood. A reduced relationship between mood in regard to HRV and/or performance could come from the poor measurement reliability of such scale. Other scales might be more appropriate for monitoring mood, such as the POMS (Wallace, Slattery & Coutts, 2014). Finally, HRV, mood states and performance are variables that can be influenced by factors unrelated to training (Ortigosa-Márquez et al., 2017), such as food and drinks, medications, stress, etc. (Rodas et al., 2008). It would be interesting to record and analyse how these factors may interact with the parameters analysed and affect training.

Future research

The aim of this research is geared towards developing a predictive system to assess training adaptations, which could help guide load distribution and detect and prevent overtraining. For such a system, and to accurately track fitness and fatigue, it is proposed to use a combination of subjective and objective variables, both from psychophysiological data and from external and internal load parameters (Haddad et al., 2017). The current study indicates a relationship between variables of internal (RPE) and external (IF) training parameters, subjective psychological states (Mood) and objective physiological data (HRV), but the relationships could be further explored. In particular, it would be of interest to look into the variations of HRV and mood at different points in time (hours and days) following training. This is because, as mentioned above, the relationship between training and HRV, or training and Mood, varies as time goes by. Tracking changes in HRV and mood in the long-term is also relevant because fitness has a longer time constant than physical and mental fatigue, meaning it asymptotes at a higher level and a later time (Morton, Fitz-Clarke & Banister, 1990), so only variations in HRV and/or mood over several weeks, rather than consecutive days, can help interpret changes in training. The study of long-term tendencies of HRV and mood as a determinant of training has already been proposed (Buchheit et al., 2010; Plews et al., 2013), but more research is still needed to consider the effectiveness of these variables for assessing training adaptation.