*1.1. Description of the Condition*

The key components in any training program are the volume (i.e., how much), intensity (i.e., how hard), and frequency (i.e., how often) of the exercise sessions, and the combination of these 'training impulses' determines the magnitude of adaptive responses that improve the physical condition of an athlete or increase fatigue [1]. Combining these key elements to optimize training in athletes for better performance represents a relevant area of research within exercise physiology and sports medicine [2]. It is recognized that a standard training program applied to a group of athletes can induce diverse responses in terms of performance and physiological adaptations [3,4]. Therefore, individualization is recognized as a training principle [1] as well as the need to adjust training stimuli to the psychophysical load capacity and individual tolerance of each athlete, if individual responses

to training and recovery loads are intended for optimal performance [5]. The maximal oxygen uptake (VO2max) is considered one of the main indicators for measuring an athlete's performance and cardiovascular adaptation to training loads [6]. The VO2max is defined as the largest volume of oxygen that the body can capture, use, and transport during intense exercise [7] and is a determining factor of endurance performance [7,8]. As Vesterinen et al. [4,9] state, although some athletes show great endurance performance improvements after standardized group training (even up to 40% in VO2max), other athletes show no changes or benefits, and sometimes even show a decrease in endurance performance. In recent years, research has looked at whether heart rate variability (HRV)-guided training has positive effects on athletic performance, given that this type of training allows daily adjustment of the training and recovery stimuli, individually based on HRV records [4,5,10].

#### *1.2. Description of the Intervention*

HRV is an indicator that enables the noninvasive analysis of autonomic nervous system activity in both its sympathetic and parasympathetic branches [11]. This is relevant if we consider that an important component of the interindividual variability in physiological responses to training is related to the balance between the parasympathetic (PNS) and sympathetic (SNS) activity of the autonomic nervous system (ANS) [12]. According to Huang et al. [13], HRV is considered the variation in the time interval between two consecutive heartbeats and obtained by calculating the time interval between two consecutive R waves (i.e., RR interval fluctuation) in the electrocardiogram (ECG). Since the elapsed time between beats is not constant, high vagally related HRV values are associated with efficient ANS, promoting behavioral adaptation and cognitive flexibility during stress [14], while low HRV is indicative of an inefficient ANS, resulting in maladaptive responses to stress and perceived threats [13]. HRV analysis is considered a useful method for measuring the heart's ability to adapt to endogenous and exogenous loads [15]; therefore, it can be used for the individual assessment of responses to training loads and recovery adaptation [4,16]. High HRV measurements indicate more parasympathetic than sympathetic activation, which is indicative of better recovery and preparedness for facing high-intensity training sessions [17]. HRV-guided training starts with a preparation period of about four weeks, which serves as a standardized data collection phase to obtain the baseline HRV values (e.g., LnrMSSD; the natural logarithm of the square root of the mean value of the sum of the squares of the differences between the adjacent RR intervals) and their normal range (upper and lower limits) for each athlete [9,18]. Once the normal range of HRV measurements has been established, the training prescribed (moderate- or high-intensity session) is based on this calculation, which is normally updated weekly [19]. Traditionally, the vagally related HRV index has been measured with ECG [20], and quantified by means of rMSSD [17]. Currently, the development and validation of new applications (i.e., smartphone applications: Kubios-HRV, Elite-HRV, Mobile Lab, or HRV4Training) facilitate daily HRV measurements and their quantification and, thus, the individual adaptation of training loads and recovery.

#### *1.3. How the Intervention Might Work*

Bellenger et al. [21], in a recent systematic review with meta-analysis, highlighted the need to use monitoring systems that accurately reflect the athletes' adaptations to the training stimulus. Although there have been numerous research studies using the HRV measure to check wellness and training adaptation in athletes [22,23], these have not focused on performance improvement based on HRV-guided training but have followed training interventions based on a traditional and nonindividualized methodology.

In contrast, evidence exists supporting the use of HRV-guided training for improved performance in endurance athletes. With this type of training monitoring, some studies have found significant VO2max improvements in athletes who have developed individualized endurance training programs based on daily HRV values. These studies alternated moderate-intensity sessions with high-intensity sessions [4,10] or even rest sessions, vigorous-intensity training, and moderate-intensity exercise [5]. However, Javaloyes et al. [18], in a program with similar characteristics developed with professional cyclists, found no significant improvements in VO2max. Likewise, significant improvements have been found among athletes following HRV-guided training in other variables; for example, for lactate in maximal test [10], speed in maximal test [4], time in maximal test [4,10], or muscle strength [24]. At the level of perceived recovery, significant improvements have also been found in variables such as general stress, emotional stress, lack of energy, and even overall mood disturbance [25]. HRV-guided training may, therefore, function as an alternative method for improving performance in resistance athletes.

#### *1.4. Why Is This Review Important?*

In the search to improve athletic performance, different training methods have been tried and studied, such as intensified training [2] or submaximal tests [26]. However, it has also been recognized that the same training program followed by a group of athletes can provoke a wide range of reactions in terms of performance and physiological adaptations [3]. Overuse injuries occur due to repetitive submaximal loading of the musculoskeletal system when there is inadequate rest to allow for structural adaptation to take place [27]. In recent years, HRV-guided training has shown itself to be a promising method for improving different performance variables (e.g., VO2max) compared to predefined training (traditional training) through the monitoring and individualization of endurance athletes' training [4,28]. HRV-guided training has been investigated in randomized trials on samples from different endurance sports, such as skiers [28], runners [4,25], and cyclists [18]), as well as athletes of different ages and levels: elite [18,28] and recreational endurance athletes [5,24,25]. Therefore, it is important to carry out a systematic review and meta-analysis of the different experimental studies conducted so far on endurance athletes in order to assess whether HRV-guided training is an effective method for performance improvement.

#### **2. Objectives**

As mentioned above, this review aimed to analyze the effect of HRV-guided training on VO2max in endurance athletes.

We asked the following research questions regarding HRV-guided training in endurance athletes:

**Research Question 1**: Does HRV-based training have an effect on VO2max? **Research Question 2**: Is the effect of this type of training superior to that of traditional training? **Research Question 3**: Is the level of the athletes decisive in obtaining an effect on the VO2max? **Research Question 4**: Does the effect of HRV-guided training determine VO2max scores according to the gender of the athlete?

#### **3. Methods**

The methods detailed below are reported in accordance with the Campbell Collaboration policies and guidelines for systematic reviews [29].

#### *3.1. Criteria for Considering Studies for This Review (Eligibility Criteria)*

#### 3.1.1. Types of Studies

We included randomized controlled trials (RCTs) and the first period of cross-over RCTs and experimental studies using a random method for the treatment assignment in order to reduce the risk of allocation bias. We restricted study eligibility by language. We did not restrict study eligibility by publication status.

#### 3.1.2. Types of Participants

We included studies with trained athletes enrolled in any form of regular endurance training (e.g., runners, triathletes, skiers, and cyclists). We included studies that recruited both men and women, or men and women separately.

#### 3.1.3. Types of Interventions

We included studies on endurance training controlled by heart rate variability to improve the athletes' performance. We considered designs comprising any dose, frequency, and duration. We also considered studies with the following types of comparisons:


#### 3.1.4. Types of Outcome Measures

	- Maximal oxygen consumption (VO2max)

#### *3.2. Search Methods to Identify the Studies*

#### 3.2.1. Electronic Searches

The register contains studies identified from the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL Complete, the Web of Science Core Collection, Global Health, Current Contents Connect, and the SciELO citation index.

The search is up to date as of 15 June 2020. The language was restricted, considering only English or Spanish. The terms used to search the databases were: (*amateur* OR *elite* OR *train*\*) AND (*HRV-guided* OR *"heart-rate variability guided"*).

#### 3.2.2. Searching Other Resources

We checked the reference lists of all the included studies and systematic reviews for additional references. We contacted experts in the field and the authors of the included studies to identify additional unpublished studies. We also checked the results of completed trials registered on the US National Institutes of Health Ongoing Trials Register, ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), and proceedings of conferences for relevant research.

#### *3.3. Data Collection and Analysis*

We conducted the following data collection and analysis in accordance with the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions [30].

#### 3.3.1. Selection of Studies

Two review authors independently screened the titles and abstracts of all the retrieved references in Microsoft Excel 2018 (Microsoft, New York, NY, USA) for Windows. The full-text study reports were retrieved for all the citations that at least one review author considered potentially relevant. Two review authors independently screened the full-text articles and identified studies for inclusion; they also identified and recorded the reasons for excluding studies in the excluded studies characteristics. Any disagreements were resolved through discussion. The selection process is detailed in a PRISMA flow diagram [31].

#### 3.3.2. Data Extraction and Management

We used a standardized piloted data collection form in Microsoft Excel 2018 for Windows and extracted the following study characteristics and outcome data: (i) Methods: study design; (ii) Participants: randomized number, study participants' mean age or age range, study location and setting, recruitment methods, inclusion and exclusion criteria, and type of endurance sport; (iii) Interventions: a description of the training intervention characteristics, the dose and duration of the training intervention, a description of the comparison intervention characteristics, the length of follow-up, the number of withdrawals, and the reasons for withdrawal; (iv) Outcomes: a description of the primary and secondary outcomes in the review that were reported in the trial and a listing of other outcomes collected in the trial; (v) Notes: the trial funding and notable conflicts of interest of the trial authors; (vi) a 'risk of bias' assessment. Two review authors independently extracted the outcome data from the included studies into Microsoft Excel 2018 spreadsheets and compared the data to identify any discrepancies in the data entries. Any disagreements were resolved by consensus. In the Characteristics of Included Studies section, we noted down if a trial did not report outcome data in a usable way. We then transferred all the outcome data into the Comprehensive Meta-Analysis software version 2.2.064 (Biostat, Englewood, NJ, USA) [32].

#### 3.3.3. Risk-of-Bias Assessment in the Included Studies

Two review authors (M.C.P., A.G.G.) independently assessed the risk of bias for each included trial using the Cochrane risk-of-bias tool [30]. Any disagreements were resolved by discussion. The risk of biases were assessed for the following domains: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment for each outcome (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other biases (such as the validity of outcome measure and baseline comparability). Each potential source of bias was assessed as either high, low, or unclear, and a quotation from the study report was provided together with a justification for the judgment in the 'risk of bias' tables. The judgments across the different studies were summarized for each of the domains listed.

#### 3.3.4. Treatment Effect Measures

The outcome data for each study were uploaded into the data tables of the Comprehensive Meta-Analysis software to calculate the treatment effects. We used the mean difference (MD) for continuous outcomes reported on the same scal, and the standardized mean difference (SMD) for continuous outcomes measured on different scales in different trials (SMD = MHRV guided training − Mcontrol group/Standard deviation) [33]. Uncertainty was expressed with 95% confidence intervals (CIs) for all the effect estimates.

#### 3.3.5. Assessment of Heterogeneity and Reporting Bias

Heterogeneity was assessed qualitatively between studies in three ways: a visual examination of the forest plots, the Chi<sup>2</sup> test (*p* <sup>≤</sup> 0.10) for heterogeneity, and the I2 statistic. The implications of the observed I<sup>2</sup> statistic value were considered as follows: 0% to 40%—might not be important; 30% to 60%—may represent moderate heterogeneity; 50% to 90%—may represent substantial heterogeneity; 75% to 100%—considerable heterogeneity [30]. Publication bias was assessed by examining the asymmetry of a funnel plot using Egger's test. If studies were distributed symmetrically around the mean effect size (ES), there was an absence of publication bias [33]. Subgroup analysis was carried out

using the outcome for athlete level (elite vs. amateur) and sex (men, women, and both sexes combined). Metaregression was used to assess the relationship between the studies and the variable sample size.

#### 3.3.6. Sensitivity Analysis

A sensitivity analysis was carried out to check whether the results varied according to the endpoint data.

#### **4. Results**

#### *4.1. Description of the Studies*

#### 4.1.1. Search Results

The search produced a total of 36 studies, with 222 additional records identified through other sources. The removal of duplicates resulted in eleven studies, which were screened by the two authors based on the title and abstract. Three studies were excluded. Eight full-text articles were assessed for eligibility. Two more studies were excluded, and six studies were included either in the qualitative analysis or in the quantitative metasynthesis. The PRISMA flow chart illustrates the search and selection process (Figure 1).

**Figure 1.** Study flow diagram following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Guidelines [31], where n is the number of papers and k is the number of individual studies.

#### 4.1.2. Included Studies

Six studies carrying out HRV-guided training with elite or amateur athletes were included in this review [4,5,10,18,19,28], which were identified by the first author and publication date: Javaloyes\_2019, Kiviniemi\_2007, Kiviniemi\_2010, Nuuttila\_2017, Schmitt\_2018, and Vesterinen\_2016.
