Authors: Asher L. Flynn, Tyler Langford, Cody Whitefoot

Department of Sport and Exercise Science, Lincoln Memorial University, Harrogate, TN, USA


Asher L. Flynn, PhD, CSCS
6965 Cumberland Gap Parkway

Asher L Flynn, PhD, CSCS is an Assistant Professor of Exercise Science at Lincoln Memorial University, TN. His research interests focus on fatigue and athlete monitoring in colligate athletes, and aspects of women’s soccer performance.

Tyler Langford, PhD, is an Assistant Professor of Exercise Science at Lincoln Memorial University, TN. His areas of research interest include exercise testing and prescription for special populations (incomplete spinal cord injury and older adults) as well as the use of effort perception for exercise prescription.

Cody Whitefoot, PT, DPT, OCS is an Assistant Professor of Exercise Science at Lincoln Memorial University, TN. His research interests include the impact of blood flow restriction (BFR) on aerobic capacity and orthopedic clinical evaluation with a special emphasis on diagnostic testing.

Reliability of a Point-of-Care Device for Saliva Analysis During Aerobic Exercise


Salivary alpha-amylase (sAA) and Cortisol (C) have been of increasing focus as biomarkers for monitoring acute and chronic stress. Recently there has been an interest in improving point-of-care devices to increase practicality of salivary testing and remove the need for laboratory analysis. A new point-of-care device has been reported to be highly reliable during baseline testing but has yet to be proven reliable up to the expected concentrations from intense exercise (exceeding 1000 U/mL). Following a modified graded exercise test (GXT) which consisted of sAA and C analysis at baseline, 50-, 70-, and 90-percent of heart rate max, the new point-of-care device showed strong to very strong reliability across the concentration spectrum (sAA: r = 0.60; C: r = 0.73). According to the results of this study, this point-of-care device is able to assist the coaching staff in making training decisions based off of the results provided.

Key Words: Alpha-Amylase, Cortisol, Athlete Monitoring, Fatigue


Salivary alpha-amylase (sAA) has been of growing interest in the sport science community as another mechanism of measuring stress. The sAA response to stress follows the sympatho-adrenomedullary (SAM) axis rather than the Hypothalamic-Pituitary-Adrenal (HPA) axis of cortisol (C), and as such has been suggested as an alternate method of monitoring stress, and this difference in response may allow for a more nuanced interpretation of stress (9, 10, 20).

Salivary alpha-amylase has been of increasing interest and many sAA responses have been investigated, namely, the response to aerobic exercise (5), resistance training (18), pain (22), social stress (15), and chronic stress (19). One of the areas of interest that is currently lacking, is to investigate the more nuanced response of aerobic stress, such as whether an intensity dependent response exists. Previous research has investigated the sAA response changes due to different intensities with variable results being reported (2), some reporting no difference (13,14,17), and others reporting significant increases in sAA concentration (6,11) as exercise intensity increases.

Another area of interest that has recently gained traction in other fields of study (e.g., psychology) is the ability for changes in sAA concentrations to help determine if an individual is in a chronic stressed state –  i.e., over trained. While this use of sAA has not made its way into the sport science realm as of yet, other research has suggested that alterations to the circadian rhythm of sAA is indicative of chronic stress (19, 21). Since monitoring athletes to reduce the risk of overtraining is a primary goal of sport science, sAA may be a useful biomarker for this.

When monitoring athletes, practicality of testing methods often requires a realistic training environment and rapid feedback. In other words, it is little use to the coaching staff if the test can only be performed in a laboratory and/or the results of a test are delayed to such an extent that decisions based on this data are no longer relevant. Since a primary focus of sport science is to reduce the chances of an athlete overtraining, rapid feedback to adjust training loads is necessary.

To address the issue of slow analysis and feedback, a number of point-of-care devices have been developed that intend to make saliva testing more practical for sport. These devices tend to have an in-mouth absorbent pad that collects saliva from a short as 30-seconds up to a few minutes, which can then be analyzed by a portable machine within a few additional minutes. The early versions of these devices lacked reliability (7,16). A newer device has been developed that has been reported to have strong reliability using professional soccer players (r = 0.93; 8) but was only used prior to a training session, when sAA concentrations would be expected to be lower. Since sAA concentrations can increase dramatically during high intensity aerobic training (pre: 314 U/ml, post: 1441 U/ml; 3), ensuring the reliability of this device through the expected range is paramount.

Therefore, the purpose of this study was twofold, 1) to determine if the Cube point-of-care device can maintain reliability during high intensity exercise, and 2) to determine if there is an increase in sAA in accordance with an increase of intensity.



Twelve participants (5 males, 7 females; 171.29 ± 8.80 cm; 71.16 ± 11.05 kg; 20.83 ± 1.95 years old) volunteered for this study. All participants had been consistently participating in moderate to vigorous physical activity for at least three-months, and most were current college athletes. All participants completed the 2022 PAR-Q+ form to ensure ability to perform high intensity exercise and were free from any injury that would affect ability to exercise on a cycle ergometer. Participants signed written informed consent document approved by the LMU Institutional Review Board prior to participation (IRB # 983).

This research project consisted of two days of testing, at least 48-hours apart. On the first day of testing, participants arrived at the lab and had anthropometric data recorded. Following collection of height, weight, and age, participants completed a standardized warm up that consisted of 5-minutes of light exercise on a stationary bike. A heart rate (HR) monitor (Polar model T31; Polar, Bethpage, New York) was used to ensure participants reached a HR of 100 bpm prior to completion of the 5-minutes. Next, participants completed a short dynamic stretch protocol, which consisted of 1 set of 10 repetitions of hamstring (leg sweeps), hip (figure 4), and quadriceps (quad pull) exercises. Following the dynamic warm-up, participants then began the graded exercise testing (GXT) protocol on the cycle ergometer.

The GXT protocol was of self-design, consisting of 2-minute stages at 70 RPMs of increasing intensity. The first stage started at 150 watts (W) and intensity was increased by 50W at each stage. Participant HR was averaged every 10 seconds (iWorx Systems Inc, Dover, New Hampshire) and recorded during the last 30 seconds of each stage. Participants were instructed to complete as many stages as possible and to attempt to achieve their calculated maximum HR (208 – 0.7*age). Upon volitional failure, resistance was removed from the cycle ergometer and participants were allowed to pedal until they felt ready to leave. During that time, researchers verbally confirmed the next testing day’s requirements (no caffeine for at least six-hours prior and no food or drink, except water, for an hour prior).

On day 2 of testing, participants arrived at the lab within an hour of their day 1 testing time in order to control for circadian rhythm, and had weight checked to ensure similar weight status as during day 1. Participant weight was checked as a proxy for hydration status. Participants verbally confirmed they had not consumed any caffeine yet that morning, nor eaten or drank anything except water for the past hour. Participants then provided their pre-exercise baseline saliva sample. The Cube device (SOMA Bioscience, Wallingford, United Kingdom) was collected first, then the under-the-tongue collection device (Salimetrics, Carlsbad, CA).

After collecting baseline saliva samples (Pre), participants completed the same warm-up as on day 1. Following the warm-up, participants completed three incremental stages (approximately 4 minutes each), using workloads associated with 50%, 70%, and 90% of their HRmax. Wattage was based on Day 1 results that elicited the target HR. Once the target HR was achieved, that HR was maintained within ± 5 bpm during the saliva collection process (approximately 3 minutes). Once both saliva samples were collected, participants moved on to the next stages (70%, 90%). When participants reached the 90% HR stage, they attempted to maintain that HR for the duration of collection, if unable to maintain that intensity, participants stopped pedaling and continued to collect saliva while sitting passively on the cycle ergometer.

As soon as the participants stopped pedaling for the 90% HRmax stage, either after completion of saliva collection, or due to inability to maintain cadence/intensity, a 10-minute timer was started to collect the recovery sAA and C timeline. Saliva samples were collected at 10-, 20-, and 30-minutes post exercise (+10, +20, +30, respectively) following the same saliva collection process as the baseline time point.

Saliva Collection and Analysis

Saliva was collected using the Cube device (SOMA Bioscience, Wallingford, United Kingdom; 30s – 2 minutes) following manufacturer’s protocol before the synthetic swab. If the participant had not collected enough saliva for the Cube device within 60s, the synthetic under-the-tongue swab (Salimetrics, Carlsbad, CA) was then added and used to collect saliva for 2-minutes. If during that time, the Cube device indicated enough saliva was collected, it was removed and preserved in the collection vial. Immediately following collection of saliva via the synthetic swab (Salimetrics, Carlsbad, CA), samples were frozen at 4°C. Samples were kept frozen until shipped overnight to the Salimetrics laboratory packaged with dry ice and ice packs to ensure samples arrived frozen. Upon receiving samples Salimetrics performed saliva analysis for sAA and C.

The point-of-care device (Cube) saliva samples were analyzed approximately 6-hours after collection following manufacturer’s protocol which consisted of dropping 2-3 drops of buffer mix onto a lateral flow device (LFD) and allowing that to incubate for 10- (single sAA analysis; 50%, 70%, 90%) or 15-minutes (dual sAA/C analysis; Pre, +10, +20, +30). After the incubation period, the cube device was used to scan the LFD and display results.

Data Analyses

In order to determine if there was an increase in sAA as intensities increased, a 1×7 Repeated Measures ANOVA, with Bonferroni adjustments as necessary, was run between all testing time points using laboratory ELISA results (Salimetrics, Carlsbad, CA). A 1×4 RM ANOVA was run to determine if there was an increase in C concentrations due to the exercise protocol. In addition, two Pearson’s correlations were run between the Cube and laboratory ELISA results across all 7 time points for sAA, and the 4 time points for C. Statistical analysis was performed using JASP (version 0.16.2). Correlation results were categorized as trivial (0.0), small (0.1), moderate (0.3), strong (0.5), very strong (0.7), nearly perfect (0.9), and perfect (1.0; 12). Cohen’s d effect sizes were categorized as trivial (0.0), small (0.2), moderate (0.6), large (1.2), very large (2.0), nearly perfect (4.0; 12).


The RM ANOVA revealed a significant difference (p < 0.001) in sAA concentration between all measurement time points (Pre, 50%, 70%, +10, +20, +30) and 90%HRmax (Figure 1). Cohen’s d effect sizes were trivial for comparisons between Pre and 50% (d = 0.16) , Pre and +30 (d = 0.03), between 50% and 70%  (d = 0.17), 50% and +30 (d = 0.19), between 70% and +10 (d = 0.07), 70% and +20 (d = 0.05), and between +10 and + 20 (d = 0.02). Cohens d effect sizes were small for comparisons between Pre and 70% (d = 0.33), Pre and +10 (d = 0.40), Pre and +20 (d = 0.38), between 50% and +10 (d = 0.24), 50% and +20 (d = 0.22), between 70% and +30 (d = 0.36), between +10 and +30 (d = 0.42), and between +20 and +30 (d = 0.41). Analysis revealed large effect sizes between all other time points and 90% (Pre, d = 1.66; 50%, d = 1.50; 70%, d = 1.33; +10, d = 1.26; +20, d = 1.28; +30, d = 1.69).

The RM ANOVA for C reported no significant differences between any time points (p = 0.82).

The correlations between the Cube device and laboratory measures were strong for sAA (r = 0.60, p = <0.001, 95% CI = 0.43 – 0.73; Figure 2) and very strong for C (r = 0. 75, p = < 0.001, 95% CI 0.59 – 0.86; Figure 3). Descriptive statistics are provided in table 1.

Figure 1:
Note. The change in sAA concentration through the testing protocol.
Data presented as means ± standard deviations.
*= Significant difference between all other time points.
sAA = Salivary alpha-amylase
Figure 2:
Note. Correlation between Point of Care device and ELISA sAA analysis.
sAA = Salivary alpha-amylase
Figure 3:
Note. Correlation between Point of Care device and ELISA Cortisol analysis.
C = Cortisol


The only significant difference being 90% HRmax compared to all other time points does provide evidence that for moderately trained individuals, while there is a trend of slight increase in sAA as intensity increases, there does seem to be an intensity threshold where the increase in sAA concentration becomes exponentially greater and this threshold may be linked to the lactate threshold (1, 4, 6). Although lactate was not measured in this study, it is assumed that participants would be above LT at 90% HRmax. Since the majority of the participants were of some type of aerobic based sport (AT; cross-country, soccer), 70% HRmax may have been too low of intensity to cross the LT/sAA threshold. The few participants that had an emphasis in resistance training (RT) did exhibit larger changes at the 70% time point than the aerobic based sports participants (RT = 99% increase, AT = 3% increase).

These results also provide evidence that sAA measurements following high-intensity training sessions may provide a more in-depth view of an athlete’s physiological stress before participating in a subsequent training session. Since there was an increase in sAA concentration following this GXT protocol, but no increase in C, along with the immediate response of sAA as opposed to the delayed response of C, provides compelling evidence that sAA may be a more appropriate method of determining the acute physiological stress of an exercise.

The new cube device has acceptable reliability with laboratory results that the data is likely useful for monitoring sAA changes due to aerobic exercise. The Cube device is also likely useful for monitoring changes in both sAA and C resting level changes. Research has indicated that changes to the baseline levels of sAA are indicative of a hypersensitive SNS system, which is a sign of long-term stress (19, 21). For example, displaced youths that lived through the Katrina catastrophe were reported to have increased resting sAA concentrations 6 months after the event compared to age matched subjects not under the stress of that event (21). War refugees that have been diagnosed with PTSD have also reported altered resting concentrations of sAA compared to those not diagnosed with PTSD (19). Therefore, this device could be useful for monitoring fluctuations in sAA concentrations among athletes so adjustments in training load can be altered to avoid chronic stress/overtraining.

As is typical with markers of chronic stress, monitoring changes in basal levels of sAA may only be useful after there are already maladaptation taking place. In order to make sAA a more useful tool to predict the decline in performance, a method of monitoring and predicting these changes is necessary. Recently this problem has been attempted to be solved by calculating the awakening response in both sAA and C (9,10). Both sAA and C follow distinct diurnal patterns within the first hour of waking. For sAA, in a healthy individual, there is a rapid and steady decline in sAA concentration from the initial waking period to 30 minutes post waking (10). Cortisol rapidly increases during the first hour after waking, followed by a rapid decline for the next hour followed by steady decline throughout the day (10). As stress begins to accumulate chronically, the awaking response for both sAA and C is altered (10, 19). Salivary alpha-amylase inverts, with lower values reported immediately upon waking and steadily increases in the first 30 minutes, while C concentrations are suppressed at both the 30-minute and hour mark (10,19). Since the new Cube device has strong correlations with both sAA and C, monitoring overtraining with athletes can be made more practical. Athletes can be taught how to take their own saliva samples, and then deliver the samples to the coaching staff in the morning. Since the sample is stabilized for approximately 24 hours in the buffer solution, the coaching staff can analyze the saliva when time allows and then make adjustments to training as soon as the same day if necessary.


This research provides more evidence that sAA does have an intensity dependent threshold before a significant increase in concentrations occur. These results also agree with previous research that the Cube point-of-care salivary analysis device provides reliable results for both sAA and C and can be used in a sport setting for saliva monitoring.


One of the primary reasons that biomarker data is not practical in a sport setting is the lack of inexpensive, reliable, and quick testing methods. The new Cube point-of-care device solves most of these issues for  sAA and C monitoring, providing the coaching staff with valuable insights into an athlete’s stressed state.

Salivary-alpha Amylase testing could be used as an alternative method for determining intensity if directly measuring lactate was not available/practical. For example, an athlete that is participating in training above lactate threshold, their time cut offs are determined prior to training, but the intensity of that prescribed speed is variable, based on the athlete’s current recovery or stressed state. The first repetition of their training day could be used as a marker to ensure the prescribed pace is eliciting the desired stress response and have training adjusted based on the results. This device could also be used for athlete monitoring from a more global stress state by monitoring the C response to training or the sAA/C awakening response, but more research is needed to determine the timeline of the changes in the awakening response due to training in order to determine the ability to predict over training.


  1. Akizuki, K., Yazaki, S., Echizenya, Y., & Ohashi, Y. (2014). Anaerobic threshold and salivary α-amylase during incremental exercise. Journal of physical therapy science, 26(7), 1059-1063.
  2. Azarbayjani, M. A., Dalvand, H., Fatolahi, H., Hoseini, S. A., Farzanegi, P., & Stannard, S. R. (2011). Responses of salivary cortisol and a-amylase to official competition. Journal of human sport and exercise, 6(2), 385-391.
  3. Bishop, N. C., Walker, G. J., Scanlon, G. A., Richards, S. T. E. P. H. E. N., & Rogers, E. (2006). Salivary IgA responses to prolonged intensive exercise following caffeine ingestion. Medicine and science in sports and exercise, 38(3), 513-519.
  4. Bocanegra, O. L., Diaz, M. M., Teixeira, R. R., Soares, S. S., & Espindola, F. S. (2012). Determination of the lactate threshold by means of salivary biomarkers: chromogranin A as novel marker of exercise intensity. European journal of applied physiology, 112(9), 3195-3203.
  5. Chennaoui, M., Bougard, C., Drogou, C., Langrume, C., Miller, C., Gomez-Merino, D., & Vergnoux, F. (2016). Stress biomarkers, mood states, and sleep during a major competition: “Success” and “failure” athlete’s profile of high-level swimmers. Frontiers in physiology, 7, 94.
  6. De Oliveira, V. N., Bessa, A., Lamounier, R. P. M. S., De Santana, M. G., De Mello, M. T., & Espindola, F. S. (2010). Changes in the salivary biomarkers induced by an effort test. International journal of sports medicine, 31(06), 377-381.
  7. Ducker, K. J., Lines, R. L., Chapman, M. T., Peeling, P., McKay, A. K., & Gucciardi, D. F. (2020). Validity and reliability evidence of a point of care assessment of salivary cortisol and α-amylase: a pre-registered study. PeerJ, 8, e8366.
  8. Dunbar, J., Hazell, G., Jehanli, A., & Wallingford, U. K. (2015, July). Investigating a dual sIgA and alpha-amylase point of care test in the sporting environment. In Abstract presented at the 12th Symposium of the International Society of Exercise and Immunology.
  9. Durguerian, A., Filaire, E., Drogou, C., Bougard, C., & Chennaoui, M. (2018). Food restriction alters salivary cortisol and α-amylase responses to a simulated weightlifting competition without significant performance modification. Journal of Sports Sciences, 36(5), 536-544.
  10. Filaire, E., Ferreira, J. P., Oliveira, M., & Massart, A. (2013). Diurnal patterns of salivary alpha-amylase and cortisol secretion in female adolescent tennis players after 16 weeks of training. Psychoneuroendocrinology, 38(7), 1122-1132.
  11. Gallina, S., Di Mauro, M., D’Amico, M. A., D’Angelo, E., Sablone, A., Di Fonso, A., … & Di Baldassarre, A. (2011). Salivary chromogranin A, but not α‐amylase, correlates with cardiovascular parameters during high‐intensity exercise. Clinical endocrinology, 75(6), 747-752.
  12. Hopkins, W. G. (2016). A New View of Statistics. Sportscience. Retrieved December 5, 2022, from
  13. Ihalainen, J. K., Schumann, M., Häkkinen, K., & Mero, A. A. (2016). Mucosal immunity and upper respiratory tract symptoms in recreational endurance runners. Applied Physiology, Nutrition, and Metabolism, 41(1), 96-102.
  14. Kunz, H., Bishop, N. C., Spielmann, G., Pistillo, M., Reed, J., Ograjsek, T., … & Simpson, R. J. (2015). Fitness level impacts salivary antimicrobial protein responses to a single bout of cycling exercise. European journal of applied physiology, 115(5), 1015-1027.
  15. Maruyama, Y., Kawano, A., Okamoto, S., Ando, T., Ishitobi, Y., Tanaka, Y., … & Akiyoshi, J. (2012). Differences in salivary alpha-amylase and cortisol responsiveness following exposure to electrical stimulation versus the Trier Social Stress Tests.
  16. Peng, H. T., Savage, E., Vartanian, O., Smith, S., Rhind, S. G., Tenn, C., & Bjamason, S. (2016). Performance evaluation of a salivary amylase biosensor for stress assessment in military field research. Journal of clinical laboratory analysis, 30(3), 223-230.
  17. Pilardeau, P., Richalet, J. P., Bouissou, P., Vaysse, J., Larmignat, P., & Boom, A. (1990). Saliva flow and composition in humans exposed to acute altitude hypoxia. European journal of applied physiology and occupational physiology, 59(6), 450-453.
  18. Souza, A. V., Giolo, J. S., Teixeira, R. R., Vilela, D. D., Peixoto, L. G., Justino, A. B., … & Espindola, F. S. (2019). Salivary and plasmatic antioxidant profile following continuous, resistance, and high-intensity interval exercise: preliminary study. Oxidative Medicine and Cellular Longevity, 2019.
  19. Thoma, M. V., Joksimovic, L., Kirschbaum, C., Wolf, J. M., & Rohleder, N. (2012). Altered salivary alpha-amylase awakening response in Bosnian War refugees with posttraumatic stress disorder. Psychoneuroendocrinology, 37(6), 810-817.
  20. Vacher, P., Filaire, E., Mourot, L., & Nicolas, M. (2019). Stress and recovery in sports: Effects on heart rate variability, cortisol, and subjective experience. International Journal of Psychophysiology, 143, 25-35.
  21. Vigil, J. M., Geary, D. C., Granger, D. A., & Flinn, M. V. (2010). Sex differences in salivary cortisol, alpha‐amylase, and psychological functioning following Hurricane Katrina. Child development, 81(4), 1228-1240.
  22. Wittwer, A., Krummenacher, P., La Marca, R., Ehlert, U., & Folkers, G. (2016). Salivary alpha-amylase correlates with subjective heat pain perception. Pain Medicine, 17(6), 1131-1136.
Print Friendly, PDF & Email