For decades, lactate threshold (LT) has been utilized by endurance athletes as a key performance predictor and to dictate training practices (29). Alongside a high V̇o2max and efficiency, a high LT is essential for endurance performance as it allows an athlete to perform at a higher work-rate for an extended duration (19). Historically, to determine LT, invasive blood measures and costly laboratory equipment are required to sample whole blood during incremental exercise tests. Furthermore, skilled technicians are required to measure blood lactate concentration values and to interpret lactate profile data. Over the past 20 years, monitoring of blood lactate concentration has become less invasive and easier to acquire because of advances in technology that require smaller amounts of whole blood and capillary blood sampling techniques. However, new alternatives to quantify exercise threshold workloads via noninvasive and cost-effective methods are constantly being introduced.
To circumvent the invasive nature of calculating exercise threshold workloads from blood lactate concentration, researchers have investigated the use of expired gas (8), and to a lesser extent heart rate (7) inflection points and rating of perceived exertion scores (10) as means for determining exercise thresholds. Although no invasive measures are required, there are still high costs related to the laboratory equipment and skilled technicians needed to collect and interpret the expired gas and heart rate data. More recently, near infrared spectroscopy (NIRS) systems have been used to continuously monitor muscle tissue oxygenation via the absorption of the near infrared light by hemoglobin and myoglobin as the near infrared light passes through the muscle tissue. Recent studies have been testing the use of the muscle tissue oxygenation values derived from NIRS to gain exercise threshold information via inflection points of muscle oxygenation parameters during incremental exercise (2,3,13,27). Although the use of NIRS is relatively new to sports science, it has proven to be a viable noninvasive tool to determine peripheral muscle oxygenation during exercise (4,22).
With further technological advancements, the use of portable NIRS systems to monitor peripheral muscle oxygenation in field testing has become more common (6). A commercially available wearable lactate threshold predicting device (WLT) (BSXinsight multisport edition; BSX Athletics, Austin, TX, USA) is a portable NIRS LED device, which is marketed as the first wearable lactate threshold predictor claiming to have the ability to determine the workload at LT during self-administered maximal cycling or running exercise tests (5). The noninvasive and portable nature of the device allows athletes and coaches to easily monitor their LT, prescribe exercise-training zones based on LT values, and monitor training adaptations in a cost-effective manner. Because of the patented design and algorithm of the device, it is difficult to evaluate the exact methodology used to predict LT workload by the WLT. However, no studies have investigated the validity nor the reliability of the WLT devices' ability to predict workload at LT. Therefore, the aims of this investigation were twofold: (a) to determine the levels of agreement between the WLT and a number of traditional blood lactate methods including the linear spline fitting method, Dmax method, modified Dmax method, fixed blood lactate concentration of 4 mmol·L−1 method, and first rise of blood lactate greater than 1 mmol·L−1 method to determine workload (km·h−1) at LT, and (b) to establish the reliability of the WLT device during a repeated trial (intradevice reliability) and when using 2 devices during a single test (interdevice reliability).
Experimental Approach to the Problem
The LT workloads calculated by the WLT were compared with traditional LT workloads after an incremental exercise test to exhaustion (GXT) on a treadmill (HP Cosmos, Traunstein, Germany). All testing was performed in controlled laboratory conditions (21 ± 1° C and 60% relative humidity). Subjects were asked to refrain from strenuous exercise (<12 hours) and to arrive at each session in a fully rested and hydrated state. For the test–retest reliability trials, subjects were to report for testing at the same time of day (±1 hour), separated by 24 hours.
Seven male (mean ± SD; age: 19–45 years, height: 172.2 ± 8.5 cm, mass: 74.5 ± 12.7 kg, V̇o2max: 58.6 ± 7.0 ml·kg−1·min−1) and 7 female (age: 18–41 years, height: 164.4 ± 7.1 cm, mass: 59.8 ± 7.6 kg, V̇o2max: 47.1 ± 6.5 ml·kg−1·min−1) subjects from a range of recreationally active to highly trained athletes (V̇o2max range = 39.7–67.0 ml·kg−1·min−1) volunteered to participate in the current study. Both male and female runners from a range of abilities and demographics were recruited to test the WLT across a large span of lactate threshold speeds (10.9–16.2 km·h−1). To be eligible for the study, all participants were required to be free from injury and medication that may have affected their ability to perform maximal exercise. Before inclusion, all participants were informed about the study including potential risks and benefits and were required to give written consent. This study was given ethical clearance by the University of Waikato Human Research Ethics Committee in accordance with the Helsinki declaration.
The GXT consisted of 3-minute stages at increasing workloads (km·h−1) determined by the WLT software. The WLT software predetermined the protocol stage workloads via information about the subjects' current fitness levels. The software required information on current “conversational pace” explained to participants as the maximal pace that subjects would still be able to comfortably maintain a conversation and “10 km pace” explained as the maximal pace subjects would be able to maintain over a 10-km distance. The GXT protocols started with a “warm up” walking stage at 4.8 km·h−1 followed by a sharp increase in workload for the second stage (9.3–11.7 km·h−1). Further stage increments ranged from 0.3 to 1.1 km·h−1 until the subject was unable to maintain the stage workload. The exercise protocol and workload increments set by the WLT had to be followed and a minimum of 4 stages had to be completed for a successful prediction of LT workload by the WLT. Expired gas variables were collected and analyzed at 15 seconds intervals throughout the GXT with an indirect calorimetry system (TrueOne 2400; Parvo Medics, Inc.; Sandy, UT, USA), which was calibrated as per the manufacturer's instructions. V̇o2max was taken as the highest V̇o2 value (ml·kg−1·min−1) recorded over a 1-minute period during the GXT.
The WLT (BSXinsight multi-sport edition; BSX Athletics) is a commercially available portable NIRS LED device that is housed in a compression sleeve and fitted over the gastrocnemius muscle of the user (Figure 1). Compression sleeves of different sizes were utilized to ensure that the compression sleeve fit firmly and comfortably on all subjects regardless of body size. During inter-reliability trials, 2 devices were used, with 1 device attached to each leg. The WLT monitors change in muscle oxygenation (total oxygenation index [TOI]) and has been suggested by the manufacturers to be able to predict LT at 97% accuracy via their patented algorithm that detects inflection points in the muscle oxygenation curve at increasing workloads (5). The WLT was aligned with the thickest section of the gastrocnemius as per the manufacturer's instructions and LT workload was calculated in pace (min·km−1), which was converted to speed (km·h−1) for further analysis.
Blood Lactate Sampling and Analysis
Blood lactate concentration was collected and analyzed (Lactate Pro 2; Arkray Global Business Inc., Shiga, Japan) in the final 10 seconds of each stage and 1-minute postexercise via finger-prick capillary blood samples using universal procedures. Traditional method LTs were calculated using a previously validated Excel spreadsheet (14,23). Traditional LT workloads were determined using a range of methods including:
Linear spline fitting (LSP): The point of intersection between 2 linear splines on a lactate curve. The location of this intersection is estimated by minimizing the sum of the squared differences between the observed lactate values and the fitted values.
Dmax: Defined as the point that yields the maximal distance from the lactate curve (using a third-order polynomial) to the line formed by the lowest and highest lactate values of the curve.
Modified Dmax (mDmax): Defined as the point that yields the maximal distance from the lactate curve (using a third order polynomial) to the line formed between 0.4 mmol·L−1 above the lowest lactate value and the highest lactate value on the curve.
Fixed blood lactate concentration of 4 mmol·L−1 (4mmoL): Calculated using an inverse prediction by finding the work rate/intensity corresponding to a lactate value equal to a fixed blood lactate concentration of 4 mmol·L−1.
First rise of blood lactate greater than 1 mmol·L−1 (1mmoL): Defined as the workload preceding an increase in lactate concentration of a fixed rise of 1 mmol·L−1 after baseline.
All LT workload calculations have also been previously described (23).
In a subset of the sample used in the current study, further trials were performed to evaluate the interdevice and intradevice reliability of the WLT (n = 12). Subjects performed 2 GXT tests separated by 24 ± 1 hour to measure the intradevice reliability of a single WLT. Given the aim was to test the reliability of the measurement device (not the test itself), the test was discontinued at exactly the same point of the GXT in the second trial. The same individuals also wore 2 WLT devices (one on each leg) during a single GXT to determine the interdevice reliability of 2 WLTs.
Descriptive statistics (mean ± SD) were calculated for all data. A 1-way ANOVA was implemented to measure differences between LT workload values derived from the WLT and traditional methods. Between-method agreements for LT were examined using intraclass correlation coefficients (ICC) with 95% confidence intervals and interpreted as 0.90–1.00 = very high correlation, 0.70–0.89 = high correlation, 0.50–0.69 = moderate correlation, 0.26–0.49 = low correlation and 0.00–0.25 = little, if any correlation (18,21). The mean differences and upper and lower levels of agreement (2 standard deviations or 95% of a normally distributed population) between methods were determined in absolute values (km·h−1) and as a percentage (%). Between-method typical error of measurement (TEM) was determined in absolute values (km·h−1) and as a percentage (%) using an excel spreadsheet (15,17). Bland Altman plots were also constructed to visualize levels of agreements. Interdevice and intradevice reliability data were log-transformed and analyzed using an Excel spreadsheet for reliability (16). Typical error of measurement and overall reliability of the WLT is presented as a coefficient of variation percentage (CV%) and as an absolute value (km·h−1) along with ICCs and upper and lower 95% CI. All statistical analyses were performed in SPSS V22.2 (IBM Corporation, Armonk, NY, USA) unless stated otherwise. Statistical significance was accepted at the p ≤ 0.05 level.
One-way ANOVA showed no statistically significant differences between the LT workloads determined by the different methods (p = 0.60). Intraclass correlation coefficients with 95% CI ranged from 0.80 to 0.98 (−0.10 to 0.99) for workload at LT calculated by the WLT and traditional methods, and are shown in Table 1.
Table 1 also shows the TEM and levels of agreement between the workloads at LT calculated by the WLT and traditional methods in absolute values and as percentages. TEM and levels of agreement and 95% CI ranged from 0.48 to 1.21 km·h−1 (3.9–10.2) and 0.18 to 0.86 km·h−1 (1.3–9.4), respectively, with the highest level of agreement between the WLT and 4mmoL method. Bland Altman plots showing the mean difference and limits of agreement for the WLT and traditional methods are shown in Figure 2.
Both interdevice and intradevice reliability resulted in highly reproducible and comparable results (Table 2). When comparing the interdevice reliability during a single test in 12 separate trials, the WLT resulted in a CV and 95% CI of 1.2% (0.9–2.1), a TEM of 0.16 km·h−1 (0.12–0.28) and an ICC of 0.97 (0.90–0.99). Similarly, the intradevice reliability determined by the same device during 2 tests separated by ∼24 hours, resulted in a CV and 95% CI of 1.2% (0.9–2.1), a TEM of 0.16 km·h−1 (0.11–0.27), and an ICC of 0.97 (0.90–0.99).
The results for the current study support both the validity and reliability of the WLT during an incremental exercise test to exhaustion on a treadmill. This study showed no significant differences in the workloads at LT determined by the WLT when compared with 5 traditional methods of measuring LT. Correlation analysis of LT speeds showed high to very high levels of agreement (ICC = 0.80–0.98) between the WLT and traditional methods, with 4mmoL threshold resulting in the highest correlation (ICC = 0.98). The typical error of measurement and levels of agreement (absolute and %) were also lowest between WLT and 4mmoL at 0.48 km·h−1 and 3.9%, and 0.18 km·h−1 and 1.3%, respectively. Additionally, the current study found high reliability when comparing 2 WLT devices during the same test (inter-reliability; CV% = 1.2, TEM = 0.16 km·h−1, ICC = 0.97), which was almost identical to the comparison between the same device during a repeated test (intrareliability; CV% = 1.2, TEM = 0.16 km·h−1, ICC = 0.97). These results suggest that the WLT is both valid and reliable in determining workload at LT through the threshold range of 10.9–16.2 km·h−1 and promote the suitability of the WLT in a practical setting.
As this is the first study to compare the WLT and traditional methods, it is difficult to make direct comparisons with existing research. However, previous studies investigating the suitability of muscle oxygenation breakpoints using other NIRS systems to determine exercise thresholds have shown similar results to those of the present study (2,3,27). It is important to note, however, that when comparing the results of the present study to existing research, the heterogeneous methodology such as the NIRS device used, site of measurement (vastus lateralis, gastrocnemius) and which parameter was used in determining the NIRS breakpoint (TOI, change in oxyhemoglobin concentration, change in deoxyhemoglobin concentration) may have an effect on the compatibility of the results. Furthermore, the majority of existing literature has compared NIRS inflection points to ventilatory thresholds (VT) (3,20,24,30) with only a handful of studies comparing NIRS inflection points to blood LTs (2,13,28). Additionally, there is a high prevalence of cycle ergometer studies in existing research which is most likely a result of the possible increase in signal noise owing to movement artifact of the NIRS leads while running (25).
The levels of agreement in our study are supported by existing literature showing high correlations between NIRS breakpoints and LTs (2,12). Similar to the results of this study, Belloti et al. (2) reported no significant differences, high correlations (r = 0.81; r = 0.76), and mean differences of 0.26 L·min−1 and 8 bpm for V̇o2 and heart rate values derived from change in deoxyhemoglobin concentration from the vastus lateralis muscle and from maximal lactate steady state during cycle ergometry. Wang et al. (28) compared threshold workload (W), V̇o2 and heart rate values from multiple NIRS parameters (TOI, change in oxyhemoglobin concentration, change in deoxyhemoglobin concentration) also from the vastus lateralis muscle with those derived from VT and LT during incremental cycling exercise. Threshold for all parameters was determined as the inflection point of 2 regression lines. The authors report the NIRS parameter with the strongest correlations to be between change in deoxyhemoglobin concentration with VT and LT (28). However, contrasting the results from our study, the authors reported a significant correlation between TOI measures and VT (r = 0.95) but not LT (r = 0.68) (28). The contrasting results between the present study and that of Wang et al. (28) could be attributed to heterogeneous methodology.
Although showing high levels of agreement through ICCs, mean difference values reveal a tendency of the WLT to overestimate LT workload when compared with traditional methods (Figure 3). However, higher workloads at NIRS inflection points compared with VT and LT has been previously reported in existing literature (11) and there is also evidence to suggest that NIRS inflection points at the gastrocnemius muscle occur at higher workloads compared with those at the vastus lateralis muscle (27). Taken together, these past findings may explain why the WLT showed a small overestimation of LT workload in our study. The WLT showed the highest levels of agreement with the 4mmoL method (ICC = 0.98; TEM = 0.48 km·h−1, 3.9%; mean difference = 0.18 km·h−1, 1.3%) with narrow 95% limits of agreement (−0.82 to 1.18 km·h−1). However, this magnitude of agreement was not apparent with all traditional methods. Visual inspection of Bland Altman plots demonstrates that outliers in are evident when comparing the WLT to Dmax and mDmax (Figures 1B,C) with data points falling outside the 95% limits of agreement. Furthermore, inspection of the scatter points for the WLT and 1mmoL (Figure 2E) also shows a downward trend with increasing workload, suggesting greater agreement between the 2 measures at higher threshold workloads. Therefore, although no significant statistical differences between the WLT and traditional methods were found, overall means and 95% limits of agreement with visual inspection of Bland Altman plots suggest that magnitude of agreements can occasionally show variation between the WLT and LT traditional methods.
Given the range of LT values that were produced in this study (12.3–13.3 km·h−1) by the different traditional methods, it makes it difficult to conclude whether the WLT provides a valid prediction of LT. Therefore, it could be argued that the reliability of the WLT is perhaps the most important factor when it comes to using the device to monitor running performance. Owing to the logistics associated with testing multiple athletes at 1 time, scientists commonly interchange multiple units of the same brand device between repeat tests. As such, it is important that the brand of device employed demonstrates good intradevice and interdevice reliability. This way, measurement errors are minimized, and a greater level of confidence is achieved when comparing and interpreting repeat tests on the same athlete. Both the intradevice and interdevice reliability of the WLT device resulted in a CV% of <1.2 (TEM < 0.16 km·h−1, ICC > 0.97), suggesting that the WLT can produce reliable and reproducible results both when using the same device and 2 separate devices. Given, it has been suggested that a CV of <10% is acceptable for reliability in other sport science measurement tools (1), such as metabolic analyzers (9), and the intradevice and interdevice CV of the one of the most commonly used blood lactate analyzers (Lactate Pro; Arkray Global Business Inc.) has been reported as 5.7 and 5.2% (26), an intradevice and interdevice CV of ∼1.2% suggests that the WLT appears to be highly reliable.
This study provides the first data on the validity and reliability of the WLT's ability to predict lactate threshold in runners. The results highlight that the WLT is a reliable tool that shows marginal variance in predicting workload at LT when compared with some of the common LT methods derived through blood-sampling techniques. These observations suggest that coaches and athletes could use this device to monitor workload at LT and prescribe training parameters based on the LT workload predictions through the noninvasive and self-administered exercise test. The nature of the exercise test prescribed by the WLT would allow coaches to remotely monitor their athletes and could allow coaches and athletes to monitor training adaptations through changes in LT workload.
The findings of our study suggest that the WLT is a practical tool for estimating lactate threshold workloads for runners. We found acceptable levels of agreement between the WLT and traditional blood lactate threshold methods and also that the device was highly reliable in both a test–retest setting and when comparing 2 devices during the same test. The WLT could be implemented by coaches and athletes as a portable and noninvasive method to monitor lactate threshold workload changes in runners.
1. Atkinson G, Nevill A, Edwards B. What is an acceptable amount of measurement error? The application of meaningful “analytical goals” to the reliability
of sports science measurements made on a ratio scale. J Sports Sci 17: 18, 1999.
2. Bellotti C, Calabria E, Capelli C, Pogliaghi S. Determination of maximal lactate steady state in healthy adults: Can NIRS help? Med Sci Sports Exerc 45: 1208–1216, 2013.
3. Bhambhani YN, Buckley SM, Susaki T. Detection of ventilatory threshold using near infrared spectroscopy in men and women. Med Sci Sports Exerc 29: 402–409, 1997.
4. Boushel R, Langberg H, Olesen J, Gonzales-Alonzo J, Bulow J, Kjaer M. Monitoring tissue oxygen availability with near infrared spectroscopy (NIRS) in health and disease. Scand J Med Sci Sports 11: 213–222, 2001.
5. BSX Athletics. A deeper look through BSXinsight—Muscle oxygenation. Available at: http://blog.bsxathletics.com/2015/02/20/muscle-oxygenation-through-bsxinsight/
. Accessed May 1, 2015.
6. Buchheit M, Bishop D, Haydar B, Nakamura FY, Ahmaidi S. Physiological responses to shuttle repeated-sprint running. Int J Sports Med 31: 402–409, 2010.
7. Bunc V, Hofmann P, Leitner H, Gaisl G. Verification of the heart rate threshold. Eur J Appl Physiol 70: 263–269, 1995.
8. Cheng B, Kuipers H, Snyder A, Keizer H, Jeukendrup A, Hesselink M. A new approach for the determination of ventilatory and lactate thresholds. Int J Sports Med 13: 518–522, 1992.
9. Crouter S, Antczak A, Hudak J, DellaValle D, Haas J. Accuracy and reliability
of the ParvoMedics TrueOne 2400 and MedGraphics VO2000 metabolic systems. Eur J Appl Physiol 98: 139–151, 2006.
10. Fabre N, Mourot L, Zerbini L, Pellegrini B, Bortolan L, Schena F. A novel approach for lactate threshold
assessment based on rating of perceived exertion. Int J Sports Physiol Perform 8: 263–270, 2013.
11. Fujimoto S, Yoshikawa T, Tateishi Y, Lixin W, Hara T, Mimura T, Nakao H, Hirata K. Evaluation of muscle oxygenation during exercise by NIRS in normal subjects - Significance of the NIRS threshold. J Jpn Col Angiol 47: 21–27, 2007.
12. Grassi B, Pogliaghi S, Rampichini S, Quaresima V, Ferrari M, Marconi C, Cerretelli P. Muscle oxygenation and pulmonary gas exchange kinetics during cycling exercise on-transitions in humans. J Appl Physiol (1985) 95: 149–158, 2003.
13. Grassi B, Quaresima V, Marconi C, Ferrari M, Cerretelli P. Blood lactate accumulation and muscle deoxygenation during incremental exercise. J Appl Physiol (1985) 87: 348–355, 1999.
14. Higgins D. Lactate-e for Microsoft Excel. 2007. Available at: http://www.uiginn.com/lactate/lactate-e.html
. Accessed March 1, 2015.
15. Hopkins WG. Analysis of validity
by linear regression. Available at: sportsci.org/resource/stats/xvalid.xls
. Accessed January 10, 2015.
16. Hopkins WG. Analysis of reliability
with a spreadsheet. Available at: http://www.sportsci.org/resource/stats/xrely.xls
. Accessed January 10, 2015.
17. Hopkins WG. A new view of statistics. Available at: http://www.sportsci.org/resource/stats/index.html
. Accessed January 10, 2015.
18. Impellizzeri FM, Marcora SM. Test validation in sport physiology: Lessons learned from clinimetrics. Int J Sports Physiol Perform 4: 269–277, 2009.
19. Joyner MJ, Coyle EF. Endurance exercise performance: The physiology of champions. J Physiol 586: 35–44, 2008.
20. Karatzanos E, Paradisis G, Zacharogiannis E, Tziortzis S, Nanas S. Assessment of ventilatory threshold using near-infrared spectroscopy on the gastrocnemius muscle during treadmill running. Int J Ind Ergon 40: 206–211, 2010.
21. Munro B. Statistical Methods for Health Care Research. Philadelphia, PA: JB lippincott, 1997.
22. Neary JP. Application of near infrared spectroscopy to exercise sports science. Can J Appl Physiol 29: 488–503, 2004.
23. Newell J, Higgins D, Madden N, Cruickshank J, Einbeck J, McMillan K, McDonald R. Software for calculating blood lactate endurance markers. J Sports Sci 25: 1403–1409, 2007.
24. Rao R, Danduran M, Loomba R, Dixon J, Hoffman G. Near-infrared spectroscopic monitoring during cardiopulmonary exercise testing detects anaerobic threshold. Pediatr Cardiol 33: 791–796, 2012.
25. Robertson FC, Douglas TS, Meintjes EM. Motion artifact removal for functional near infrared spectroscopy: A comparison of methods. IEEE Trans Biomed Eng 57: 1377–1387, 2010.
26. Tanner R, Fuller K, Ross MR. Evaluation of three portable blood lactate analysers: Lactate Pro, lactate scout and lactate plus. Eur J Appl Physiol 109: 551–559, 2010.
27. Wang B, Xu G, Tian Q, Sun J, Sun B, Zhang L, Luo Q, Gong H. Differences between the vastus lateralis and gastrocnemius lateralis in the assessment ability of breakpoints of muscle oxygenation for aerobic capacity indices during an incremental cycling exercise. J Sports Sci Med 11: 606–613, 2012.
28. Wang L, Yoshikawa T, Hara T, Nakao H, Suzuki T, Fujimoto S. Which common NIRS variable reflects muscle estimated lactate threshold
most closely? Appl Physiol Nutr Metab 31: 612–620, 2006.
29. Yoshida T, Chida M, Ichioka M, Suda Y. Blood lactate parameters related to aerobic capacity and endurance performance
. Eur J Appl Physiol 56: 7–11, 1987.
30. Zhang Z, Wang B, Nie Q, Luo Q, Gong H. Portable muscle oxygenation monitor based on near infrared spectroscopy. Front Optoelectron China 2: 248–252, 2009.