Association of Novel Oral Appliance Use With Heart Rate Variability in Adolescent Athletes

Introduction

Heart rate variability (HRV) analysis offers a noninvasive approach for assessing autonomic nervous system function. Since 1996, numerous clinical applications of HRV analysis have been documented [1], making it a valuable tool in sports science for monitoring individual athletic performance [2,3]. A previous systematic review suggested that HRV biofeedback is an effective, safe, easy-to-learn, and easy-to-apply method for athletes to enhance sports performance [4]. Moreover, analysis of HRV data collected during physical exercise has substantial potential in sports physiology [5]. For athletes, coaches, and scientists alike, optimizing athletic performance remains a primary focus. Previous research has highlighted various factors that influence athletic performance, including physical conditioning [6], nutrition [7], sufficient sleep, and adequate recovery [8,9]. Another approach to enhancing athletic performance involves the use of oral appliances (OAs) [10]; custom-made OAs have demonstrated positive results [11].

OAs are typically classified into 2 main categories: mandibular advancement devices (MADs) and tongue-repositioning or tongue-retaining devices for stabilization [12]. Various types of OAs are available on the market and generally fall into 3 major categories: MADs, soft palatal lift devices, and tongue-stabilizing devices [13]. MADs include mandibular advancement splints and mandibular repositioning devices [14]. MADs help reduce snoring and treat obstructive sleep apnea (OSA) by decreasing overjet, mandibular intercanine distance, and lower arch crowding [15]. In a clinical trial examining the efficacy of MADs for individuals with snoring, MADs effectively reduced palatal flutter and tongue-base snoring [16]. However, the impact of OAs on HRV in athletes has received limited attention.

A previous study suggested that HRV might be a useful indicator for assessing the efficacy of MAD therapy (a type of oral appliance) in patients with OSA [17]. Another study showed that 3 months of treatment with an oral jaw-positioning appliance led to an increase in high-frequency (HF) power (0.15–0.4 Hz) of HRV, indicating enhanced parasympathetic activity. Additionally, the ratio of low-frequency (LF) power (0.04–0.15 Hz) to HF power (ie, LF/HF ratio) decreased, reflecting reduced sympathetic activity in patients with mild OSA [18]. In the present study, we introduced a novel patented oral appliance, the Lin Oral Appliance (LOA), which features a tongue compressor. This device is characterized as an airflow-interference-type, nasal-congestion-relieving, snore-reducing oral appliance. Patent numbers for the LOA are I602555 (Taiwan) and ZL-2013-1-0753192.9 (China) [19]. LOA braces were affixed along the entire dental arch of the upper jaw, and the tongue compressor was positioned behind the center of the upper jaw. One edge of the tongue compressor was secured between the first and second molars on the left side of the mandible; the other edge was anchored between the first and second molars on the right side of the mandible. Typically, the tongue compressor applies pressure to the center of the tongue, facilitating unobstructed airflow to the back of the throat. This mechanism helps prevent obstruction by the posterior oral cavity and tongue, thus expanding the oral airway during breathing. Additionally, support from the tongue compressor may promote coordination of the fascial system, which surrounds and connects organs, muscles, and other structures, thereby facilitating integrated whole-body function [20]. Because our previous study demonstrated that continuous use of the LOA led to a mean reduction of 0.02% per day in snoring rate among adults [21], the present study explored the impact of the LOA on HRV in adolescent athletes.

Material and Methods

STUDY DESIGN AND PARTICIPANTS:

This observational study was conducted to explore the impact of the LOA on HRV in adolescent athletes. The study protocol was reviewed and approved by the Institutional Review Board of the Research Ethics Committee of Buddhist Dalin Tzu Chi Hospital, Taiwan (No. B11002009).

This study enrolled 43 healthy adolescent athletes from a professional karate school – without specific restrictions on body mass index (BMI) – who had been practicing karate for an average of 2.7 years. They had been provided with the LOA at a dental clinic approximately 6 months before the study. Additionally, the participants did not take any medications during the study period. Participation was voluntary; prior to enrollment, written informed consent was obtained from all participants and their legal guardians because the adolescent athletes were senior school students under 20 years of age. All athletes wore the LOA and underwent HRV testing at their respective schools under a coach’s supervision.

MEASURE:

The LOA is constructed from medical-grade ethylene/vinyl acetate and consists of 2 components: custom-made dental braces and a tongue compressor with adjustable length. The tongue compressor can be set to various lengths, including 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, or 3.5 cm. For participants in this study, a length of 2.5 cm was used (Figure 1).

To mitigate the potential influence of time-of-day effects [22], all test sessions were conducted in the afternoon, specifically between 2: 00 and 4: 00 PM, to minimize the impact of daily physiological variation and sleep-wake status on HRV. Participants were instructed to follow a regular, balanced diet without high-calorie foods and to refrain from consuming caffeinated beverages within 4 hours prior to the test [23]. However, pretest activity was not objectively quantified, which we acknowledge as a limitation. Future studies should incorporate objective external and/or internal load metrics (eg, session rating of perceived exertion, training duration, or wearable-derived training load) to better control for the acute effects of recent training. HRV measurements were obtained from the athletes on 4 occasions at fixed times after meals while they were seated with their eyes closed in an environment with the temperature controlled at 27°C. To assess the impact of the LOA on HRV, each test session lasted 10 minutes, resulting in a total testing duration of 40 minutes. Each participant underwent HRV testing 4 times under the following conditions: before wearing the LOA for 10 minutes (test 1), while wearing the LOA for 10 minutes (test 2), immediately after removing the LOA (test 3), and 10 to 20 minutes after removing the LOA (test 4). The study design is illustrated in Figure 2.

Before placement on the xiphoid process of the sternum, the electrodes of a Polar H10 device (Polar Electro Oy, Kempele, Finland) were moistened with water at room temperature. The chest strap was then secured around the participant’s chest immediately below the pectoral muscles [24]. A 10-minute electrocardiography recording was subsequently obtained while the participant remained in the seated position, both with and without the LOA. The electrocardiography unit of the Polar H10 device records cardiac electrical signals at a frequency of 1000 Hz and has been validated against 3-lead electrocardiography, which is considered the gold standard in this field [25]. Polar Sensor Logger is an application that enables logging of RR intervals via Bluetooth connectivity. This application uses the Polar Software Development Kit, which allows access to raw data for analysis and processing [26].

HRV is a noninvasive method used to assess dynamic autonomic nervous system activity through measurements of neurocardiac function [27]. HRV analysis includes linear and nonlinear methods. Linear methods are further categorized into time-domain and frequency-domain analyses. Time-domain indices include the proportion of adjacent RR intervals differing by more than 50 milliseconds (p.NN50), the standard deviation of differences between adjacent RR intervals (SDSD), and the square root of the mean of the squares of differences between adjacent RR intervals (RMSSD). Frequency-domain indices include HF power (0.15–0.4 Hz), LF power (0.04–0.15 Hz), and the LF/HF ratio. To reduce potential confusion in selecting parameters that reflect sympathetic and parasympathetic activity, further analysis was conducted to assess correlations among parameters in both the time and frequency domains. This analysis aimed to identify parameters that could serve as reliable indicators of sympathetic and parasympathetic activity.

In a previous study, the LF/HF ratio was identified as an indicator of sympathovagal balance [28]. In the present study, we used the LF/HF ratio, a frequency-domain index, as a surrogate for sympathetic activity, in accordance with methodologies previously described by AlQatari et al [29] and Kobayashi et al [30]. Furthermore, RMSSD was used as a surrogate measure of parasympathetic activity [31,32].

STATISTICAL ANALYSES:

Electrocardiogram data were downloaded, then analyzed using R software version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria) with the RHRV package. Data files were subjected to visual inspection for artifacts, and necessary corrections were applied manually or via software. Time- and frequency-domain analyses were performed to evaluate HRV. Correlation tests were conducted separately for variables in the time and frequency domains. Frequency-domain analysis evaluates variability in HF and LF components, reflecting the activity of distinct branches of the autonomic nervous system. If the assumption of sphericity for repeated measurements, as determined by Mauchly’s test, was not met, the P value was adjusted using the Greenhouse-Geisser epsilon correction for comparisons between groups. Bonferroni correction was then applied for post hoc analyses. Two-way repeated-measures analysis of variance (ANOVA) was used to examine the associations of sex and test sequence with HRV. Independent variables listed in Table 1 were included in a multiple linear regression model for repeated measures. This approach typically involves the use of mixed-effects models or generalized estimating equations (GEEs) when the data do not follow a normal distribution. Before analysis in R software, the entire 40-minute testing duration for each participant was segmented into test periods (tests 1–4), and HRV indices for each period were calculated using RR intervals. All statistical analyses were 2-sided, and the statistical significance threshold was set at P=0.05.

Results

Data from 43 of the 46 participants were included in the analysis conducted between August 1, 2021, and May 31, 2022. Three participants were excluded due to missing data. The mean participant age was 14.8±1.6 years. Baseline characteristics did not significantly differ between male and female participants. Table 1 presents the demographic characteristics of the participants. Participants had been training for an average of 2.7±1.6 years and spent approximately 10 hours per week practicing karate.

We conducted further analyses to explore correlations among parameters in both the time and frequency domains. Variables within the time domain and the frequency domain showed significant correlations with each other, as demonstrated in Table 2. One-way repeated-measures ANOVA revealed significant differences in the mean RMSSD, p.NN50, and SDSD across the 4 tests (P=0.002, P<0.001, and P=0.001, respectively), as shown in Table 1. Further analysis via 2-way repeated-measures ANOVA indicated significant sex differences in p.NN50 (P=0.019), LF/HF ratio (P=0.002), and LF (P=0.008) (Table 3). Post hoc comparisons with Bonferroni correction revealed significant differences in RMSSD and SDSD between test 1 and test 4. The p.NN50 showed significant differences concerning test 1 vs test 3, test 1 vs test 4, and test 2 vs test 4. Significant differences in LF power and the LF/HF ratio were also observed between test 2 and test 3 (Table 4). During these tests, RMSSD significantly increased from test 1 (before LOA use) to test 4 (10–20 minutes after LOA removal). Post hoc tests with Bonferroni correction showed a mean increase of 12.2 milliseconds (P=0.007) (Figure 3). The within-participant effect size for the increase in RMSSD, estimated using Cohen’s d, was 0.533, indicating a moderate positive effect. GEEs adjusted for autocorrelation showed that test round (10 minutes per round) was a significant predictor (estimate=4.07, P<0.001, α=0.877) of RMSSD. In contrast, sex, BMI, age, and LOA were not significant in the GEE model adjusted for autocorrelation. For the LF/HF ratio, none of these factors displayed statistical significance in the GEE model (α=0.574) (Table 5). Nonetheless, post hoc analysis revealed a significant difference in the LF/HF ratio between test 2 and test 3 (P=0.033) (Figure 4).

Discussion

In the GEE model, LOA use during test 2 was not significantly associated with RMSSD, whereas test round was positively associated with RMSSD. Moreover, the within-participant effect size (Cohen’s d=0.533) appeared larger than the RMSSD increase typically attributable to seated rest alone. A previous meta-analysis suggested that prolonged sitting is not associated with significant changes in HRV, including RMSSD [33]. Therefore, we speculate that the RMSSD elevations observed in tests 3 and 4 reflect a carryover effect from LOA use in test 2. The estimated within-participant correlation was strong (α=0.877), supporting the appropriateness of considering repeated measures in the GEE framework. After adjustments for sex, age, and BMI, adolescent athletes demonstrated a significant increase in parasympathetic activity from test 1 to test 4, which may be related to LOA use in test 2. In contrast, sympathetic activity did not significantly change from test 1 to test 4, as determined by the LF/HF ratio in a multivariable GEE model with autocorrelation.

In the present study, we observed that RMSSD, SDSD, and p.NN50, which are time-domain parameters, can serve as proxies for parasympathetic activity. This interpretation is supported by the strong correlations observed among these time-domain parameters. Notably, RMSSD and SDSD exhibited the strongest correlation, exceeding 0.832. Based on these findings, we propose that RMSSD, SDSD, or p.NN50 be used as surrogate measures within time-domain analysis for assessing parasympathetic activity. Notably, previous research has suggested that SDNN and the SDNN index are the most sensitive HRV parameters for assessing the autonomic response to rectal distention in patients with irritable bowel syndrome, indicating that different physiological contexts may require distinct HRV parameters to accurately reflect autonomic nervous system activity [34]. RMSSD, SDSD, p.NN50, SDNN, and the SDNN index were positively correlated with each other, suggesting that these parameters are similarly indicative of parasympathetic activity. Therefore, although RMSSD showed promise as a surrogate measure of parasympathetic activity in our study population of adolescent athletes, further research is needed to validate its utility across diverse populations and physiological conditions. Additionally, comparative studies evaluating the performance of RMSSD against other HRV parameters in various contexts would help clarify its specific strengths and limitations.

Among frequency-domain parameters, the LF/HF ratio was significantly positively correlated with LF and negatively correlated with HF. Based on the observed correlation coefficient of 0.339 (P<0.001) between LF/HF and LF – and consistent with previous studies – we propose using the LF/HF ratio as an indicator of sympathetic activity [29,30]. However, limitations associated with interpreting the LF/HF ratio should be considered, including potential confounding factors and interindividual variability. Future studies should further evaluate the validity and utility of the LF/HF ratio across diverse populations and physiological contexts.

Overall, HRV reflects the capacity of the autonomic nervous system to respond to various physiological and environmental stimuli, including breathing patterns, physical exercise, mental stress, hemodynamic and metabolic fluctuations, changes in sleep and posture, and compensatory mechanisms in response to illness and disease [35,36]. HRV also provides important insight into the relationship between stress and health [36]. It is widely used as a noninvasive tool for assessing autonomic nervous system function across numerous clinical conditions, such as atherosclerotic progression, congestive heart failure, diabetic neuropathy, and depression [37]. HRV analysis also is commonly used to examine the relationship between athletic performance and recovery [38].

In this study, we used time-domain (specifically RMSSD) and frequency-domain (specifically the LF/HF ratio) indicators as markers of parasympathetic and sympathetic activity, respectively. Our findings demonstrated that adolescent athletes wearing the LOA exhibited significantly increased parasympathetic activity. This result is consistent with previous research indicating that time-domain measures, such as RMSSD, are more consistent than frequency-domain measures for describing chronic cardiovascular autonomic adaptation in athletes [39]. Another study showed that self-myofascial release can enhance the recovery process, potentially by increasing parasympathetic activity [40]. Previous studies have also suggested that heightened parasympathetic activity contributes to improved emotional regulation [41]. A plausible mechanism is engagement of the well-established trigeminocardiac reflex [42,43]. Mechanical stimulation of the oral cavity (teeth, palate, and tongue) can activate trigeminal afferents projecting to the reticular formation, thereby increasing vagal efferent output from the dorsal motor nucleus and nucleus ambiguus to the heart and other viscera, which results in enhanced parasympathetic HRV indices. Another potential mechanism is that the LOA improves upper airway patency and breathing stability [44], while attenuating sympathetic overactivity and restoring autonomic balance [45]. However, the impact of autonomic nervous system function on athletic performance warrants further investigation.

Mouthguards are dental devices used by athletes to protect the teeth, tongue, gums, and cheeks from injury during contact sports. Relative to mouthguards, OAs differ in material (soft copolymer), structure (custom-designed by a dentist or sleep physician), and purpose. OAs are recommended not only for adult patients with primary snoring but also for those with OSA who cannot tolerate continuous positive airway pressure therapy or who prefer an alternative treatment [46]. We previously demonstrated that the LOA significantly reduced snoring frequency [47]. In the present study, we observed that adolescent athletes wearing the LOA exhibited increased parasympathetic activity. However, further studies are needed to clarify the potential benefits of OAs for enhancing athletic performance and overall well-being.

Prior studies have shown that autonomic function declines with age in both sexes and that sex-related differences in parasympathetic regulation diminish after approximately 50 years of age [48,49]. In adults, HRV has also shown inverse associations with BMI [50,51]. However, we observed no significant associations of HRV with age, sex, or BMI in our cohort of adolescent athletes. This discrepancy might be explained by the participants’ young age and the relatively narrow BMI distribution, which could have restricted variability and reduced statistical power to detect such relationships. These considerations limit the generalizability of our findings and underscore the need for studies in broader age groups and BMI ranges, as well as nonathlete populations. Further work is also required to clarify the long-term effects of LOA use in athletes.

The LF/HF ratio is commonly used as a marker of sympathovagal balance; higher values are often interpreted as reflecting sympathetic predominance, whereas lower values suggest parasympathetic predominance. However, this interpretation has been questioned by multiple authors [28,52,53]. In particular, LF power (0.04–0.15 Hz) reflects a composite of sympathetic and vagal influences that is strongly affected by baroreflex dynamics, whereas HF power (0.15–0.40 Hz) largely reflects respiratory sinus arrhythmia. The LF/HF ratio also is sensitive to mean heart rate, which further limits its utility as a standalone index of sympathovagal balance, especially in clinical or free-living settings. For a more robust characterization of autonomic regulation, it is preferable to report absolute LF power and absolute HF power along with time-domain metrics (eg, RMSSD) and – when appropriate – nonlinear indices. Accordingly, an important consideration in the present study is that we used the LF/HF ratio as a proxy for sympathetic activity; our inferences based on this metric should be interpreted with appropriate caution.

This study has several limitations that should be acknowledged. First, the study design lacked randomization because participants volunteered to wear the LOA under the supervision of their legal guardians. This aspect may have introduced selection bias and evaluation bias. Second, this observational, nonrandomized, within-participant study lacked both a sham, time-matched control condition and a counterbalanced test order; therefore, time- and sequence-related factors during the prolonged resting measurements should be considered when interpreting the HRV findings. Additionally, pretest training load was not objectively quantified, and respiratory patterns were not monitored – both of these factors may influence HRV indices. Third, all participants breathed spontaneously during the recording period. Given that breathing rate and depth can substantially affect HRV, standardized paced breathing or controlled respiratory conditions might have yielded different results; accordingly, the use of free breathing represents a methodological limitation. Fourth, although participants had been training for an average of 2.7 years, the relationship between training duration and HRV-related outcomes requires further investigation. In test 4, potential time-related or carryover effects of the LOA on parasympathetic HRV indices could not be rigorously evaluated within the short time frame. Fifth, the absence of information concerning test-environment humidity, menstrual status in female participants, and objective verification of caffeine abstinence indicates the need for future studies that include more detailed demographic and physiological data about adolescent athletes. Sixth, pretest activity was not objectively quantified. Future studies should incorporate objective external and/or internal training-load measures (eg, session rating of perceived exertion, training duration, or wearable-derived training load) to better control for the effects of recent training on baseline HRV. Finally, all tests were conducted within a single 40-minute session; thus, the long-term effects of wearing the LOA in athletes warrant further investigation.

Conclusions

Adolescent athletes showed a significant increase in parasympathetic activity from test 1 to test 4 after adjustments for sex, age, and BMI. This increase may be associated with LOA use during test 2. Future studies should incorporate sham devices, randomized and counterbalanced protocols, and concurrent respiratory monitoring to clarify causality and underlying mechanisms.