Daily Resting Heart Rate Variability in Adolescent Swimmers during 11 Weeks of Training

dolescent athletes are particularly vulnerable to stress. The current study aimed to monitor one of the most popular and accessible stress markers, heart rate variability (HRV), and its associations with training load and sleep duration in young swimmers during an 11-week training period to evaluate its relevance as a tool for monitoring overtraining. National-level swimmers (n=22, age 14.3 1.0 years) of sprint and middle distance events followed individually structured training programs prescribed by their swimming coach with the main intention of preparing for the national championships. HRV after awakening, during sleep and training were recorded daily. There was a consistent ~4.5% reduction in HRV after 3–5 consecutive days of high (>6 km/day) swimming volume, and an inverse relationship of HRV with large (>7.0 km/day) shifts in total training load (r= 0.35,p<0.05). Day-to-day HRV did not signi cantly correlate with training volume or sleep duration. Taken together, these ndings suggest that the value of HRV uctuations in estimating the balance between the magnitude of a young athlete's physical load and their tolerance is limited on a day-to-day basis, while under sharply increased or extended training load the lower HRV becomes an important indicator of potential overtraining. Keywords: autonomic nervous system; competitive swimming; high-intensity exercise; sleep; training volume 1. Introduction For training to be e ective, it is essential to have both appropriate planning and to implement a training load

and their tolerance is limited on a day-to-day basis, while under sharply increased or extended training load the lower HRV becomes an important indicator of potential overtraining. Keywords: autonomic nervous system; competitive swimming; high-intensity exercise; sleep; training volume 1. Introduction For training to be e ective, it is essential to have both appropriate planning and to implement a training load monitoring and adjustment system that includes biofeedback [1]. Sport practitioners are constantly looking for ways to most objectively, quickly, and cost-e ectively design individualized training programs. Such programs are based on the individual adaptation, as re ected by the recuperation (restitution) and super-compensation of the body functions during the recovery period after training sessions, to maximize long-term gains in performance while at the same time avoiding excessive fatigue, overtraining and injuries. For a successful design of such programs, it is important to use integrated training monitoring tools that allow recording variables representing changes in whole-body functional state, rather than uctuations of very speci c/isolated markers. One marker of the functional status of an organism, that gives an indication of the balance between the sympathetic and parasympathetic nervous system, is heart rate variability (HRV). This indicator has recently Int. J. Environ. Res. Public Health2020,17, 2097; doi:10.3390 /ijerph17062097 /journal/ijerph

Int. J. Environ. Res. Public Health2020,17, 2097 2 of 12 become quite popular and has been associated with numerous conditions ranging from sleep quality to sports performance [2–6]. HRV represents uctuations in electrocardiogram (ECG) R–R time interval and re ects cardiac regulation by the autonomic nervous system based on the instantaneous sum e ects of sympathetic and parasympathetic systems in response to both physical and psychological stimuli [7]. Exercise activates the sympathetic nervous system, causing an increase in myocardial contractility and vasoconstriction of peripheral blood vessels, which are moderated by the parasympathetic nervous system [8,9]. Recent reviews suggest that HRV biofeedback is a method that is e ective and safe, and is easy to learn and apply to improve sport performance [4,8]. It has also been highlighted that HRV may re ect the training-induced level of stress and recovery, which has encouraged regular HRV monitoring [3,10]. However, it has also been shown that overload training leading to overreaching has little e ect on resting HRV or increased postexercise HRV, and that day-to-day load variability was not related to HRV (for review, see Bellinger et al., [3]). Thus, although HRV has been investigated extensively, its practical use in everyday training remains controversial [2,3]. Handling individual training-induced responses is particularly relevant in such sports as swimming, where the risk of overreaching and overtraining is high because of the very large volume and monotonous training loads and early specialization by athletes [11]. As reviewed by Koenig et al. [12], although HRV monitoring in swimming may have some bene cial outcomes, it is also recognized that there is a lack of translational approaches to apply the current evidence to general practice. Several recent studies demonstrated the relationships between the autonomic nervous system, performance, and fatigue, showing that HRV responses appeared to depend on the training status, age, and environmental factors of the athletes [13–17]. However, the vastly di erent populations and contexts of these studies create uncertainty about the interpretation of their ndings, which may partly explain why sport practitioners are still reluctant to incorporate HRV monitoring into their arsenal of training tools. Because

fatigue, showing that HRV responses appeared to depend on the training status, age, and environmental factors of the athletes [13–17]. However, the vastly di erent populations and contexts of these studies create uncertainty about the interpretation of their ndings, which may partly explain why sport practitioners are still reluctant to incorporate HRV monitoring into their arsenal of training tools. Because of the above considerations, we feel that the utility of HRV as a training load monitoring tool requires further examination in young competitive swimmers. Adolescents were selected for the investigation because they may be particularly vulnerable to overreaching because of their structural, biomechanical, and, consequently, functional alterations related to the intrinsic e ects of rapid somatic growth and development [18,19]. Adolescence is also associated with increased sensitivity to environmental conditions (stress related to school activities, relationships with parents, friends, etc.) and therefore it is particularly important to control stress and recovery in competitive athletes of this age. Better understanding of training-speci c HRV responses may optimize athletic activity and prevent the development of long-lasting fatigue. The main aim of this study was therefore to analyze HRV associations with training load and sleep duration during the 11-week training macrocycle of young swimmers. We hypothesized that HRV measurements would be feasible during the preparatory training period of young swimmers, that total training volume and high-intensity (HI) training volume would be inversely correlated with HRV, and that sleep duration would be directly correlated with HRV. 2. Materials and Methods 2.1. Participants Twenty- ve adolescent swimmers at national level from the same swimming school were initially recruited to the study. The inclusion criteria were: (1) healthy, (2) absence of any clinical history of neuromuscular disorders within preceding year, (3) regular swimming training of at least 5 years; (4) competitive at least at national level, and (5) early or middle adolescence (13–16 years). However, three athletes were excluded because of inconsistent R–R interval measurements. The characteristics of the participants for whom data were analyzed (n=22) are shown in Table. Athletes were following periodized training programs that were individually developed and prescribed by their

least 5 years; (4) competitive at least at national level, and (5) early or middle adolescence (13–16 years). However, three athletes were excluded because of inconsistent R–R interval measurements. The characteristics of the participants for whom data were analyzed (n=22) are shown in Table. Athletes were following periodized training programs that were individually developed and prescribed by their

Int. J. Environ. Res. Public Health2020,17, 2097 3 of 12 swimming coach with the main intention of preparing them for the national swimming championships. This period of training had been preceded by a nonathletic period of passive rest during the summer holiday and a transition period from rest to regular training that comprised light training 4–5 times per week for two weeks. A standard training week comprised 5–7 swimming sessions, most of which were held in the afternoons on Mondays to Fridays (the days they all attended lessons in the school during the morning hours), and some sessions in the mornings during the week. Just before each afternoon swimming session, participants performed dry-land mobility and strength exercises of 30–45 min in duration. Training required a minimum of 8 hours per week. There was a substantial variation in training compliance (76.0 17.2%), but the athletes with low compliance (two below 50%) were not excluded from the analyses because load variability was considered as a factor in HRV. Swimmers specialized in events of di erent strokes at short to middle distances and had at least 5 years of swimming training experience. All participants and their parents signed an informed consent form prior to participation. The study was approved by the relevant bioethics committee of the Lithuanian Sports University (No. BEK-KIN(B)-2019-184). The study was conducted in accordance with the Declaration of Helsinki. Table 1.Characteristics of the participants (mean and standard deviation). Variable Male (n =7) Female (n =15) Total (n =22) Age, years 15.4 (0.7) 13.8 (0.6) ˆ 14.3 (1.0) Height, cm 179.5 (6.0) 165.1 (6.7) ˆ 169.7 (9.3) Weight, kg 65.5 (6.7) 56.5 (6.1) ˆ 59.4 (7.5) Body fat, % 10.8 (3.9) 21.5 (3.8) ˆ 18.1 (6.3) Knee extension peak torque, Nm/s 197.3 (20.7) 141.4 (27.5) ˆ 159.2 (36.6) Vertical jump height, cm 40.2 (2.1) 31.2 (3.3) ˆ 33.8 (5.1) VO 2peak, mL/min/kg 47.8 (5.0) 38.8 (6.8) ˆ 41.6 (7.5) Maturity Tanner II, n (%) 1 (14.3 %) 2 (13.3 %) 3 (13.6 %) Tanner III, n (%) 5 (71.4 %) 11 (73.4 %) 16 (72.8 %) Tanner IV, n (%) 1

197.3 (20.7) 141.4 (27.5) ˆ 159.2 (36.6) Vertical jump height, cm 40.2 (2.1) 31.2 (3.3) ˆ 33.8 (5.1) VO 2peak, mL/min/kg 47.8 (5.0) 38.8 (6.8) ˆ 41.6 (7.5) Maturity Tanner II, n (%) 1 (14.3 %) 2 (13.3 %) 3 (13.6 %) Tanner III, n (%) 5 (71.4 %) 11 (73.4 %) 16 (72.8 %) Tanner IV, n (%) 1 (14.3 %) 2 (13.3 %) 3 (13.6 %) Note: ˆp<0.05 vs. male. 2.2. Study Design The swimmers were followed for 77 days of the preseason training period (September to November) of which 11 days were preplanned to be without training sessions. HRV and sleep duration were measured daily by the athletes themselves. The intensity and volume of the training load of each swimmer were recorded daily by the swimming coaches in the athlete's training log and then provided to the researchers. The training loads were not adjusted or corrected as a result of the data collected in the framework of the current study, and decisions to select training loads, modalities, intensities etc., were completely at the discretion of the athletes' coaches. During the period of the study, the swimmers took part in 2–3 local competitions. One week before the experimental period, participants were familiarized with the use and proper tting of heart rate and sleep monitoring devices, their somatic maturational status was veri ed by experienced medical sta using the Tanner scale [20]. We evaluated participants' body height (Anthropometry Martin, GPM Siber-Hegner, Geneva, Switzerland), body weight and fat percentage (Tanita, model TBF 300; Tokyo, Japan), jump height (Power Timer Testing System, Newest, Finland), dominant leg knee extension strength at angular speed 30 s 1 (System 3; Biodex Medical Systems, Shirley, NY, USA), and VO2peakon a stationary cycling ergometer (Ergoline, Windhagen, Germany) using a portable breath-by-breath analyzer (Oxygen Mobile; Jaeger/VIASYS Healthcare, Hoechberg, Germany).

Int. J. Environ. Res. Public Health2020,17, 2097 4 of 12 2.3. Training Load Monitoring The speci c swimming training was performed in a 25-m standard swimming pool and was strati ed according to heart rate (HR) into intensity zones 1 to 4:<139 beats/min (bpm) for Zone 1, 150 10 bpm for Zone 2, 170 10 bpm for Zone 3, and>181 bpm for Zone 4; sprints with maximal exertion level were classi ed as Zone 5. The intensity control was based on criterion speed individually directed by coach in conjunction with pulse count using a chronometer “Alpha” (Sport-Thieme, Grasleben, Germany) for 15 s after distinct swimming tasks. All swimmers were experienced in manually measuring pulse. Swimming at Zones 4 and 5 was considered HI training. Total swimming volume and HI swimming volume were analyzed as distance covered during the training period. 2.4. HRV Measurement Over the 11 consecutive weeks of their preparatory training period, swimmers daily measured their HRV in the supine position for 2 min immediately after awakening in the morning by positioning an H10 Bluetooth HR strap (Polar Electro, Kempele, Finland) paired with a freely available smartphone application (Elite HRV, Ashville, NC, USA) on their chest. This system has been used previously for daily measurement of HRV [21]. We analyzed the square root of the mean sum of the squared di erences between R–R intervals (RMSSD), which was converted by logarithmic transformation (lnRMSSD) to avoid outliers and simplify the analyses, as suggested by Nakamura et al. [22]. Data les were visually inspected for artefacts, and corrections made manually if necessary. Both RMSSD and lnRMSSD are recognized markers of parasympathetic activity and are the preferred HRV markers for eld-based monitoring [2]. 2.5. Sleep Monitoring During the study, all participants slept in their homes and went to bed at around 11 pm. Participants wore an activity monitoring bracelet on the wrist of their nondominant hand every night. Daily sleep duration was monitored by a commercially available wrist-worn sleep/activity tracker, Mi band 2 [23]. Moderate intraclass correlation coe cients (ICC) of 0.62–0.75 have been reported for the Mi Band 2 sleep duration

in their homes and went to bed at around 11 pm. Participants wore an activity monitoring bracelet on the wrist of their nondominant hand every night. Daily sleep duration was monitored by a commercially available wrist-worn sleep/activity tracker, Mi band 2 [23]. Moderate intraclass correlation coe cients (ICC) of 0.62–0.75 have been reported for the Mi Band 2 sleep duration assessment [24]. 2.6. Statistical Analyses All variables are expressed in terms of mean standard deviation (SD). Levene's test was used to test the homogeneity of variances. The Kolmogorov–Smirnov test was used for checking distribution normality. A one-way analysis of variance (ANOVA) was used to determine the e ects of time on dependent values (load, HI load, sleep, RMSSD and lmRMSSD) across weeks or days during the training period. If a signi cant main e ect was found, the signi cance of the di erence between means was estimated by applying pairedt-tests. Statistical power (observed power, OP) was calculated and presented where appropriate. During the monitoring period, there were 23 cases when individual swimmers had 3–5 consecutive days of high total training volume (>6 km/day), and 24 cases of 3–5 consecutive days of passive rest not related to injuries or illnesses (on several occasions for some individuals). These cases were analyzed separately by comparing HRV at the beginning and end of each period using pairedt-tests. An independentt-test was used for comparisons of baseline values between females and males. The level of signi cance was set at 5% (p<0.05). Pearson's correlation (r) was used to quantify the relationship between each daily RMSSD or lnRMSSD and the previous day's training variables. The correlation was considered strong ifr>0.5; moderate ifr=0.3–0.5; and weak if r=0.1–0.3 [25]. All data analyses were performed using IBM SPSS Statistics software (v.22; IBM Corp., Armonk, NY, USA).

Int. J. Environ. Res. Public Health2020,17, 2097 5 of 12 3. Results Body weight, height, and dry-land performance were higher in males than in females and fat percentage was lower in males than in females (bothp<0.05) (Table). Because HRV and maturation level were not gender dependent, all the data were pooled for the analysis. 3.1. Training Volume Total swimming training volume was 232.1 81.7 km during the 11 weeks, with training at HI (combined zones 4 and 5) comprising 6.7 4% of total swimming volume. There was a signi cant week-by-week variation in total swimming volume, with the highest peak at week 6 (>6.0 km/per day, p<0.05, OP=0.99) (Figure). Peaks of the swimming volume that reached HI zones were evident in weeks 4, 8, and 9 (about 0.5 km/per day,p<0.05, OP=0.65). Total training volume was reduced on Wednesdays (p<0.05 compared with Mondays and Tuesdays, OP=0.95) and Saturdays (p<0.05 compared with any other day, OP=0.99), with usually no load on Sundays. HI swimming volume did not vary signi cantly between days within a week.Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 5 of 12 day, p < 0.05, OP = 0.99) (Figure 1). Peaks of the swimming volume that reached HI zones were evident in weeks 4, 8, and 9 (about 0.5 km/per day, p < 0.05, OP = 0.65). Total training volume was reduced on Wednesdays (p < 0.05 compared with Mondays and Tuesdays, OP = 0.95) and Saturdays (p < 0.05 compared with any other day, OP = 0.99), with usually no load on Sundays. HI swimming volume did not vary significantly between days within a week. Figure 1. Total (a, b) and high-intensity (c, d) training volume, sleep (e, f) and lnRMSSD (g, h) in swimmers across the training period (average ± SD). Note: # p < 0.05 vs. previous value. 3.2. Sleep Average duration of sleep ranged from 7.32 ± 1.04 to 8.73 ± 0.69 h per day between subjects, with no change in weekly sleep duration over the training period (Figure 1). Within a week, swimmers were getting less sleep on Mondays

in swimmers across the training period (average ± SD). Note: # p < 0.05 vs. previous value. 3.2. Sleep Average duration of sleep ranged from 7.32 ± 1.04 to 8.73 ± 0.69 h per day between subjects, with no change in weekly sleep duration over the training period (Figure 1). Within a week, swimmers were getting less sleep on Mondays compared with any other day (p < 0.05 in all cases, OP = 0.99) and particularly compared with Sundays (>1 h, p < 0.05). 3.3. Resting HRV (lnRMSSD) During the training period, the group average values of HR, RMSSD, and lnRMSSD were 68.6 ± 6.9 bpm, 73.0 ± 24.7 ms, and 4.23 ± 0.35 ms, respectively. Weekly variation in resting lnRMSSD reached significance in weeks 5 and 6 (p < 0.05 compared with the previous week) (Figure 1) but was generally minimal and usually did not coincide with training volume peaks. Interestingly, morning lnRMSSD showed a progressive decrease within each week from a peak on Sundays and Mondays to the lowest values on Fridays (p < 0.05, OP = 0.98). In addition, lnRMSSD decreased from 4.52 ± 1.91 to 4.32 ± 1.73 ms after 3–5 days of high total training volume (> 6 km/day), and increased from 4.09 ± 1.63 to 4.29 ± 1.54 ms after 3–5 days of rest (p < 0.05 in both cases). Figure 1. Total (a,b) and high-intensity (c,d) training volume, sleep (e,f) and lnRMSSD (g,h) in swimmers across the training period (average SD). Note: #p<0.05 vs. previous value. 3.2. Sleep Average duration of sleep ranged from 7.32 1.04 to 8.73 0.69 h per day between subjects, with no change in weekly sleep duration over the training period (Figure). Within a week, swimmers were getting less sleep on Mondays compared with any other day (p<0.05 in all cases, OP=0.99) and particularly compared with Sundays (>1 h,p<0.05).

Int. J. Environ. Res. Public Health2020,17, 2097 6 of 12 3.3. Resting HRV (lnRMSSD) During the training period, the group average values of HR, RMSSD, and lnRMSSD were 68.6 6.9 bpm , 73.0 24.7 ms, and 4.23 0.35 ms, respectively. Weekly variation in resting lnRMSSD reached signi cance in weeks 5 and 6 (p<0.05 compared with the previous week) (Figure) but was generally minimal and usually did not coincide with training volume peaks. Interestingly, morning lnRMSSD showed a progressive decrease within each week from a peak on Sundays and Mondays to the lowest values on Fridays (p<0.05, OP=0.98). In addition, lnRMSSD decreased from4.52 1.91 to 4.32 1.73 ms after 3–5 days of high total training volume (>6 km/day), and increased from4.09 1.63 to 4.29 1.54 ms after 3–5 days of rest (p<0.05 in both cases). 3.4. Correlations between Daily HRV, Training Volume, and Sleep Quantity Individual day-to-day lnRMSSD values varied up to 50% across the training period (p<0.05) (Table). There was a weak inverse correlation (r> 0.10) of lnRMSSD with total swimming volume for 14 of the 22 subjects, which reached signi cance in only ve swimmers (r> 0.27,p<0.05). The correlations of lnRMSSD day-to-day variability with HI training volume or sleep volume were even more trivial and unpredictably shifted from positive to inverse depending on the subject. A moderate inverse relationship was detected between lnRMSSD and large shifts in total training volume (r= 0.35, p<0.05 in cases of a>7.0 km/day swimming volume increase/decrease) (Figure). No signi cant correlations were detected between lnRMSSD and small or moderate uctuations (increase/decrease by>3.0 or>5.0 km/day) in total volume or HI training volume. Table 2. Individual values of R–R intervals (RMSSD) converted by logarithmic transformation (lnRMSSD) across the training period and correlations (Correl) with total and high-intensity (HI) training and sleep duration. LnRMSDD, ms Correl LnRMSDD No Gender Average Max Min SD CV% Training Volume HI Volume Sleep 1 M 4.30 4.91 3.61 0.29 6.7 0.05 0.02 0.01 2 M 4.25 4.88 2.20 0.36 8.5 0.22 0.24 0.13 3 M 4.53 4.99 4.04 0.21 4.6 0.07 0.03 0.13 4 M 3.90 5.19 3.18 0.44 11.3

This research investigates HRV in adolescent swimmers over an 11-week training period.