Rugby Union requires players to maintain high performance levels and imposes substantial physical workloads for prolonged periods of time across the playing season (Quarrie et al., 2017). To accomplish optimum results, it is necessary to develop a range of player capacities (e.g. low to high intensity running, strength training.) and skills (e.g. passing, kicking, wrestling) in a combination that may vary according to playing position, competitive level and match situation (Austin et al., 2011). In this context, a knowledge of the demands of the game (i.e. internal and external workload and energy expenditure (EE)) it is necessary to customize training practices and nutritional strategies, for optimal body composition and performance, for specific positional roles and competitive levels (Fontana et al., 2015; 2016).

The internal and external workload of rugby has been studied using different methodological approaches like the session-rating of perceived exertion (RPE), heart rate (HR) monitors and global positioning systems (GPS) (Cunniffe et al., 2009; Halson, 2014; Quarrie et al., 2017). On the contrary, few studies have focused on EE in rugby players: overall daily EE in rugby players was estimated using SenseWear armbands (Bradley, et al. 2015a; 2015b) or calculating the EE of accelerated running by using GPS in Rugby League players (Kempton et al., 2015). However, possibly due to the practical and technical limitations of the above methods in field conditions (Morehen et al., 2016), the EE of actual playing/training in Rugby Union is still unreported.

Indirect calorimetry, based on the direct measure of oxygen uptake (VO₂), is the most common method to determine EE in exercise laboratories (Jequier et al, 1987). In addition, based on the well-known linear relationship between heart rate (HR) and VO₂, alternative indirect approaches for the VO₂ estimate and the subsequent estimate of EE have been developed (Åstrand et al., 2003). These approaches allow accurate and precise estimates of EE in a variety of activities (Achten and Jeukendrup, 2003) and have been successfully applied in studies of elite Rugby Union (Da Lozzo and Pogliaghi, 2013); however, their practical applicability is reduced by the need of a preliminary incremental test in the laboratory (to establish the individual HR/VO₂ relationship), that increases the overall financial cost and reduces the time-efficiency of the approach.

In 2011 Wicks and colleagues developed a simple HR index (HR_index) method to estimate VO₂ in healthy non-athletes and clinical populations without the need for a laboratory test to determine the individual HR/VO₂ relationship. The method was based on the observation that a valid linear relationship exists between HR_index (calculated as actual HR/resting HR) and VO₂ expressed in multiples of the resting metabolic rate (VO_2METs) (Wicks et al., 2011). Within the well-known limitations related to the use of HR measurement for VO₂ estimation (Achten and Jeukendrup, 2003), the HR_index method may provide a new, low cost and easy to use alternative to existing methods to estimate VO₂ and EE during Rugby activities. In fact, HR monitors are relatively “cheap” equipment, already used by elite clubs and may also represent a convenient investment for lower level clubs offering important information on the absolute and relative intensity of exercise to a wide range of coaches and athletes. Furthermore, HR_index has been investigated in its applicability in predicting aerobic fitness (VO_2max) (Esco et al., 2012; Haller et al., 2013), a possibility that could be useful to determine and monitor players’ fitness levels without requiring expensive metabolic-carts.

However, this appealing field method is based on a linear HR_index/VO₂ relationship that has only been validated in a non-athletic population. Consequently, the application of HR_index to individuals with a high fitness level requires preliminary validity verification.

Within this background, the current study tested the following hypotheses: i) a valid linear relationship between HR_index/VO_2METs can be confirmed in rugby players; ii) VO₂ and EE of incremental treadmill running can be accurately and precisely estimated based on HR_index in rugby players; iii) players’ VO_2max can be estimated using HR_index.

Fifteen professional rugby players (24 ± 3 years, 8 forwards: 106 ± 10 kg, 1.89 ± 0.09 m, 18 ± 8 % fat mass; 7 backs: 93 ± 11 kg, 1.81 ± 0.08 m, 14 ± 6 % fat mass) playing in the Italian Rugby Union First Division, were recruited and included in the study after they gave their written, informed consent. The study was approved by the Departmental Ethics Committee and all experiments were done in accordance with the principles laid down in the Declaration of Helsinki. All participants were non-smokers, free of any musculoskeletal, respiratory, cardiovascular and metabolic conditions that may influence the physiological responses during exercise testing. All the incremental tests were conducted in the exercise physiology laboratory of the sports sciences section of the University of Verona on an electromagnetically controlled treadmill (Runrace, Technogym, Italy).

Participants were instructed to avoid any physical exercise in the 48 hours before the testing session. In the morning, after medical clearance and after a 10-min rest in a seated position, HR was measured for a period of 3 min, using a heart rate monitor (Polar Electro Oy, Finland, acquisition frequency: 5 seconds). Resting HR was identified as the lowest HR of the three minutes of monitoring (Haller et al., 2013). Then, body mass (digital scale, Seca 877, Seca, Leicester, UK), height (vertical stadiometer, Seca, Leicester, UK) and percentage of body fat (based on the sum of the 6-skinfold thickness (Fontana et al., 2015)) were determined.

On the same day, participants performed a step-wise incremental test consisting of 1-minute baseline measurements, 3 minutes of warm-up at 8.0 km/h, followed by step-wise increments of 0.5 km/h every minute until volitional exhaustion. The treadmill was set at 1% gradient to replicate outdoor over-ground running (Jones and Doust, 1996). The above ramp incremental protocol was chosen to obtain a time to exhaustion around 8-12 minutes (American College of Sports Medicine, 2017) using a method extensively described elsewhere (Pogliaghi et al., 2014). In brief, based on previous data by our group and others (Duthie et al., 2003; Pogliaghi and De Roia, 2007) we anticipated VO_2max values ranging between 46 and 52 ml·kg^-1·min^-1; considering the following energy cost of running (American College of Sports Medicine, 2017):

Breath-by-breath pulmonary gas exchange and ventilation were continuously measured using a metabolic cart (Quark b², Cosmed, Italy) as previously described (De Roia et al., 2012). Oxygen uptake (VO₂⁾, HR, and EE were calculated for each 1-minute step as average of last 10 second of each step. Maximal VO₂ (VO_2max) and maximal HR (HR_max) were determined as the highest values reached upon exhaustion.

After verification of Wick’s relationship between METs and HR_index (see below in “statistical analysis”), individual HR_index was calculated as actual HR/resting HR and the Wick equation was applied to estimate a 5-s average VO₂ for each step during the incremental test and to estimate VO_2max (Wicks et al., 2011):

Finally, energy expenditure (EE-HR_index), relative to time (kcal·min^-1) was calculated for every step of the incremental exercise based on an energy equivalent for VO₂ equal to 5.0 kcal·L^-1 (di Prampero, 1986).

Data are presented as means ± standard deviation throughout. After preliminary assumptions were verified, the linear relationship between HR_index and METs, was modelled and Pearson’s product moment correlation coefficient was calculated. The original Wicks equation and equation derived from this sample of rugby players were compared through the difference between R² (Field, 2013). Furthermore, differences among mean values of directly measured and estimated VO₂ and between measured and estimated EE were compared by two-way Repeated Measures ANOVA (method of estimate and speed), followed by Holm-Sidak correction. In addition, directly measured and the predicted values of VO_2max were compared using a paired t-test.

The correlation between measured and estimated submaximal VO₂, VO_2max, and between measured and estimated EE were also modelled and Pearson’s product moment correlation coefficient was calculated. Lastly, a Bland-Altman analysis followed by one-sample z test was used to determine accuracy, and precision of the measured and estimated VO₂, EE, and VO_2max values. All statistical analyses were performed using SigmaPlot 11.0 (Systat Software Inc) and α = 0.05; statistical significance was accepted when p < α. 95% Confidence Intervals (CI) were also reported (Cumming, 2014).

The mean resting value of HR was 61 ± 5 b/min. All the subjects completed the incremental test with no adverse events (time to exhaustion= 12.0 ± 2.83 minutes, maximal speed = 14 ± 1.4 km/h, VO_2max = 47.1 ± 4.3 ml·kg^-1·min^-1). HR, VO₂, and EE values increased as a function of speed during the incremental test. Submaximal and maximal data are presented in Table 1.

A strong positive and linear relationship was found between HR_index and METs (METs = 5.89 x HR_index – 4.88, R²= 0.92, Figure 1), almost identical to that reported by Wicks in sedentary individuals (METs = (HR_index x 6) – 5, R²= 0.95). The application of our equation did not significantly improve the prediction R² compared to the original equation reported by Wicks (ΔR² = -0.00069, p < 0.001). Therefore, the original Wicks equation was used for the estimation of VO₂ and EE.

Repeated measures ANOVA showed a significant main effect of speed on the submaximal values of VO₂ and EE (p < 0.001 and p < 0.001 for VO₂ and EE respectively). On the contrary, there was no significant effect of method (measured vs. predicted) (p = 0.17; p = 0.16) on submaximal VO₂ and EE values (Figure 2 panels A and B). Furthermore, measured and estimated values of VO₂ (ml·kg^-1·min^-1) and measured and estimated values of EE were highly correlated (VO₂: Figure 2 panel C, r = 0.96, p = 0.00, 95% CI [0.95, 0.97]; EE: Figure 2 panel D, r = 0.94, p = 0.00 95% CI [0.92, 0.95]). In addition, the Bland-Altman analysis confirmed a non-significant bias and a relatively small imprecision between measured and estimated VO₂ (Figure 2 panel E, bias = 0.60 ml·kg^-1·min^-1, z = 0.52, precision = 2.89 ml·kg^-1·min^-1) and between measured and estimated EE (Figure 2 panel F, bias = 0.00 kcal·kg^-1·min^-1, z = 0.04, precision = 0.02 kcal·kg^-1·min^-1). Finally, estimated VO_2max was not significantly different from the measured value (47.1 ± 4.3 vs. 46.8 ± 4.3 ml·kg^-1·min^-1, p = 0.91, 95% CI [44.0, 47.9]). Furthermore, measured and estimated values were highly correlated (r = 0.96) and with a non-significant bias (bias = 0.17, z = 0.22) and good precision (precision = 1.29 ml·kg^-1·min^-1).

With the aim to validate, in an athletic population, the use of a “field” method developed by Wicks et al. (2011) in sedentary individuals, we tested the hypothesis that VO₂, EE, and VO_2max can be accurately predicted using HR_index in rugby players during incremental treadmill running. The study confirmed a valid linear relationship between HR_index/VO_2METs in well-trained professional rugby players, that was not different from that previously validated in non-athletic adults (Wicks et al., 2011). In agreement with our hypothesis, the predicted VO₂, EE, and VO_2max were not significantly different from the direct measures, and highly correlated with them. The validity of the HR_index-based approach in this population offers the opportunity to estimate submaximal VO₂ and EE from simple HR measures, sparing the requirement of a preliminary test to determine the individual HR-VO₂ relationship. Furthermore, HR_index could offer new and inexpensive alternatives to the existing indirect methods (e.g. 20-m shuttle run test) for the assessment and periodic monitoring of athletes’ aerobic fitness (VO_2max).

The measurement of internal load and EE is particularly important in rugby, where the individualization of training and recovery strategies could be challenging as a consequence of player numbers, size and heterogeneity, the variability of movement tasks and the high training loads of the game (Fontana et al., 2015; Quarrie et al., 2013). However, the existing methods to estimate EE present limitations: the gold standard method, i.e. the doubly labelled water technique (Schoeller et al., 1986), has been used in elite Rugby League (Morehen et al., 2016) to measure EE, providing a good estimation of the EE during a certain “time window”. Still, this method is impractical to track short-term variations of EE and, when used to monitor daily or weekly EE, it is prone to error due to the confounding effect of non-rugby activities and to the possible inaccuracy in recording food consumption (Gropper and Smith, 2013). In addition, the high financial costs, the need for specific equipment and trained staff make this method inaccessible to all but the top-level elite clubs.

Match analysis and global positioning systems devices have also been used to estimate EE (Cummins et al., 2017; Kempton et al., 2015). However, these techniques may lead to possible EE underestimations due to the inability to account for static efforts (e.g. scrums, mauls) as well as short-distance running that represent the primary movement characteristics of players in the forward role during a rugby match (Buchheit et al., 2015; Dubois et al., 2017; Quarrie et al., 2013). Dubois et al. (2017) recently compared the metabolic demands of professional Rugby Union using these two methods (i.e. match analysis and GPS) and HR monitors, finding that both match analysis and GPS were unable to detect 40% of the HR-detected exertion (Dubois et al. 2017). SenseWear Armbands were also used to assess EE in Rugby Union during off-season and on-season periods (Bradley, et al. 2015a; 2015b), but their use is limited since they cannot be worn during matches, physical contact or during bad weather conditions. Finally, as for the double labelled water technique, both GPS and Armbands may be affordable/manageable by top level clubs only.

The HR_index approach allows sports scientists to estimate EE, overcoming most of the limitations of the above methods; furthermore, it eliminates the requirement of the preliminary identification of the individual HR/VO₂ relationship, inherent to previously applied HR-based methods (Da Lozzo and Pogliaghi, 2013).

In the present investigation we found that the approach originally developed by Wicks et al. (2011), in non-athletic populations, retains its validity when applied to well-trained rugby players. The fact that the anthropometric (body mass, height and % body fat) and functional characteristics (VO_2max) of the rugby players tested in our study were comparable to those of their international counterparts (Duthie et al., 2003a; Fontana et al., 2015). suggests the generalizability of our findings to First Division players; however, further investigation in these populations is warranted.

The use of HR to estimate and monitor EE during intermittent exercise is sometimes questioned due to: i) inability of HR to account for the anaerobic component of exercise, potentially causing an underestimation of energy expenditure of highly intense, intermittent activities; ii) a relatively slow response to changes of exercise intensity (i.e. a relatively slow “on-phase”, mirrored by a slow “off-phase” when intensity decreases, similar to the time-course of VO₂) (Miyamoto and Niizeki, 1992); iii) a disproportionately high response of HR at very low exercise intensities (Achten and Jeukendrup, 2003). In addition, coaches and researchers should be aware of the effect of environmental factors and within subject variation on the HR/VO₂ relationship during activity: hydration status, fatigue and increases in core temperature (Achten and Jeukendrup, 2003) may lead to non-metabolic increases of HR that could bias VO₂ and EE estimation during very long training/game sessions and in the summer period.

Within these limitations, HR has been shown to be a valuable method to the track physiological responses in field conditions during different activities ranging from soccer (Alexandre et al., 2012) to a military loaded run (Colosio and Pogliaghi, 2018); furthermore, no differences were found between HR response during field or laboratory tests performed by either professional or amateur soccer players (Chamari et al., 2005; Hoff et al., 2002). These findings support the overall validity of HR monitoring to estimate VO₂ and EE during outdoor activities/team sports under ecological conditions.

The third objective of this investigation was to compare directly measured values of VO_2max with the values predicted using HR_index. Similar to previous studies (Esco et al., 2012; Haller et al., 2013; Wicks and Oldridge, 2016), the average predicted VO_2max (47.1 ± 4.3 ml·kg^-1·min^-1) was not significantly different from the measured VO_2max (46.8 ± 4.3 ml·kg^-1·min^-1, p = 0.91) and they were highly correlated (r = 0.96). This finding supports the general applicability of HR_index to investigate the mean fitness level of a group of subjects (or team). Furthermore, and in contrast with the existing literature, (Esco et al., 2012; Haller et al., 2013), our study demonstrated a good agreement between the predicted and measured values of VO_2max (mean bias between values = 0.17 ml·kg^-1·min^-1, z = 0.22, precision = 1.29 ml·kg^-1·min^-1). While these results seem promising, repeatability of the method remains to be determined.

It should be noticed that the ramp incremental protocol used in this investigation was chosen to respond to two main requirements: i) the validation of the HR_index/VO₂ relationship across a wide range of intensities (from low to maximal effort) without including some of the confounding factors that could be present in the most common field tests (e.g. changes of direction, terrain characteristics) ii) to obtain a time to exhaustion around 8-12 minutes and by doing so, an accurate VO_2max (American College of Sports Medicine, 2017). Regarding the second point, our time to exhaustion was slightly higher than expected (12.0 ± 2.83 minutes). However, it has been shown that, VO_2max is unaffected by speed increments in trained subjects (Kang et al. 2001; Sperlich et al. 2015). The above data support the validity of the measured values for VO_2max of the present study.

Finally, considering the laboratory environment of this investigation and the limitations of HR measurements, further studies should investigate HR_index applicability outside the laboratory environment, during training/game and field testing sessions.

We validated the use of the HR_index, originally developed by Wicks et al. (2011) on a non-athletic population, to estimate VO₂, EE and VO_2max of treadmill running in professional rugby players. The main advantages of HR_index in comparison to existing methods are: i) low costs and minimal staff training requirements ii) time efficiency iii) the possibility to obtain time-resolved EE estimates with daily frequency iv) the applicability during matches/training. Finally, HR_index appears to provide accurate estimates of VO_2max in this population of professional rugby players. Within the limitation of HR-based methods, and even if further studies are needed to validate EE estimates deriving from HR_index in ecological conditions, this low cost and easy-to-use field method could give insight into the demands of the game in different positional roles and competitive levels, informing the customization of training practices and nutritional strategies.