Journal Logo


The Gas Sampling Interval Effect on V˙O2peak Is Independent of Exercise Protocol


Author Information
Medicine & Science in Sports & Exercise: September 2017 - Volume 49 - Issue 9 - p 1911-1916
doi: 10.1249/MSS.0000000000001301
  • Free


The measurement of maximal oxygen consumption (V˙O2max) is a key variable during both sports performance-based and clinically based cardiopulmonary exercise (CPX) tests. CPX testing has become more common and useful as advancements in computer technology have made V˙O2 measurement less burdensome. Of particular importance is the fact that current technology allows technicians to choose from a variety of gas sampling intervals, including short (e.g., breath-by-breath or 5-s block averages), moderate (e.g., 10- to 15-s block averages), and long-duration intervals (e.g., 30- to 60-s block averages), to obtain measures of V˙O2 through the continuum of rest to maximal effort.

Although it is often the intent to measure V˙O2max during CPX testing, the value reported as V˙O2max might more accurately be called V˙O2peak, as it typically reflects the highest value achieved during testing, which may or may not represent the cardiorespiratory system's true maximal rate. Research signifying that the gas sampling interval can significantly alter the V˙O2peak value stands as evidence that values may not be deemed V˙O2max with certainty (1,23). In particular, the longer sampling intervals tend to produce lower values while shorter sampling intervals often result in relatively higher values. These differences may be due to the increased variability that is introduced as sampling intervals are shortened (19). One tradeoff that must be considered is whether the lengthening of the sampling interval to limit variability unintentionally leads to underestimation of the true V˙O2max (20). This potential misclassification of V˙O2max could result in a misguided clinical decision regarding therapy or faulty disease prognosis. For athletes, small differences in V˙O2max could provide misinformation regarding the success of a training program and erroneously alter future training strategies. In an effort to reduce variability and more accurately identify V˙O2max, both general (“any averages larger than breath-by-breath but smaller than 60 s” [11]) and specific (“15-breath moving averages” [20], “8-breath average” [18], and “30-s averages” [4]) sampling recommendations have been made. However, there is currently no consensus on the most appropriate gas sampling interval to use, resulting in a variety of intervals used across different protocols and populations within the literature. In addition, many researchers fail to specify which interval they have chosen.

Studies that have examined gas sampling intervals have used a variety of CPX protocols to elicit V˙O2peak (e.g., treadmill ramp [11,18], treadmill staged [16], and cycle staged [1,3]). Exercise modality is known to alter the achieved V˙O2peak (15,22). To the authors' knowledge, only a single study has investigated V˙O2peak using different sampling intervals under varying protocols (3); alas, no comparison of gas sampling intervals between protocols was completed. In as much as the protocol and gas sampling interval can independently alter CPX variables, interaction effects do not appear to be established and presented in the literature.

Self-paced V˙O2max (SPV) protocols are a relatively new form of CPX testing (14). SPV allows participants to alter work rate, at their discretion, in an attempt to maintain a particular effort, based on Borg's RPE scale (5). With this protocol, the potential for constantly varying work rate may reduce the likelihood of a metabolic steady state as efforts toward maximal effort occur. An SPV's potential absence of steady-state work rates may warrant the need for an altered gas sampling interval to correctly measure V˙O2peak, without the undue influence that variability has on the under- or overestimation of the true V˙O2max. Sampling intervals chosen in recent SPV studies have included 30-s block averages (13,14,21), 15-breath moving averages (8,10), and 15-s block averages (2).

In light of the various gas sampling intervals used throughout the CPX literature with different protocols and populations, three main goals were identified for this study to assess the role of protocol and subject population on the gas sampling interval effect on V˙O2peak. The first goal was to ascertain the differences in measured V˙O2peak during an SPV as a function of the deliberate manipulation of the gas sampling interval. The second goal was to determine whether differences that exist between sampling intervals remain when a Bruce protocol, rather than an SPV protocol, is used in the same subjects. The third and final goal was to confirm our findings in a second, independent subject sample. We hypothesized that manipulating the gas sampling intervals would alter the measured V˙O2peak during an SPV, that the change in V˙O2peak between sampling intervals would produce a similar pattern when comparing the SPV and Bruce protocols, and that differences in V˙O2peak between sampling intervals would be of similar magnitude when compared with differences in an independent subject sample.



This study used data sets from two different research studies that were completed in our laboratories. Each study was approved by a university Human Subjects Institutional Review Board. All subjects were considered “low risk” based on the risk stratification guidelines of the American College of Sports Medicine and completed an informed consent before testing procedures. Data from the first study were used to investigate the effect of the intentional manipulation of gas sampling intervals on V˙O2peak during the SPV protocol. Also, the same data were used to investigate the effect on V˙O2peak during the Bruce protocol. Thirteen participants (eight males and five females; 24 ± 3 yr; 75.9 ± 9.5 kg; 175.3 ± 7.6 cm; V˙O2peak via 15-breath moving averages, 56.2 ± 6.8 mL·kg−1·min−1) were recruited for the first study (here forward known as the first subject sample). These subjects regularly completed ≥150 min·wk−1 of moderate-intensity exercise. Finally, data from the second study were used to confirm our findings with regard to the fact that gas sampling interval affects V˙O2peak independent of exercise protocol. For this independent sample comparison, the second study recruited seven participants (two males and five females; 21.9 ± 2.7 yr; 57.7 ± 7.7 kg; 168.7 ± 7.9 cm; V˙O2peak via 15-breath moving averages, 54.2 ± 8.0 mL·kg−1·min−1) (second subject sample). These subjects completed an average of 4 h·wk−1 of running.


First subject sample

In the first study, subjects completed a Bruce protocol treadmill exercise test, and then on a subsequent visit, they completed an SPV on the same treadmill (8). The main finding was that V˙O2peak values between the Bruce and the SPV protocols were not different (8). Bruce protocol tests consisted of standard speeds and inclines consistent with the protocol starting at 1.7 mph and 10% incline, increasing speed and incline every 3 min (6). In the initial research, subjects exercised until volitional exhaustion, and a verification phase was completed after 10 min of recovery to confirm whether the effort was indeed maximal. Also, a familiarization to the SPV was completed to mitigate the likelihood that the maximal values associated with SPV would be influenced by a lack of familiarity. The SPV protocol consisted of walking and running on the treadmill at an incline of 8%. The self-chosen speed was selected by the participant to achieve the prescribed RPE levels of 11, 13, 15, 17, and 20—each for 2 min (14). The RPE scale was made visible and referenced verbally to remind participants of the effort level for each stage.

Second subject sample

The second set of data came from subjects completing CPX tests as part of a simulated altitude study (7). Each CPX test started as two 4-min running economy stages and then transitioned to a more traditional maximal protocol. At familiarization, the subjects transitioned to the third stage of the Bruce protocol after which standard speeds and inclines were followed (6). The subjects exercised until volitional exhaustion. With an established V˙O2peak, the CPX protocols at baseline and the follow-ups consisted of two 4-min running economy stages and then transitioned to a modified Astrand protocol with subsequent increases in the incline of 2% every 2 min until volitional fatigue.

Instrumentation and Measures

For both subject samples, gas sampling was completed using a True One 2400 metabolic cart (ParvoMedics, Sandy, UT), and exercise tests were completed on either a TrackMaster TMX425C or TMX428CP treadmill (Full Vision, Inc., Newton, KS). The metabolic cart was flow calibrated before each testing session using a 3-L syringe and manufacturer instructions. Gas calibration consisted of a two-point calibration using room air and a known gas quantity (16% O2, 4% CO2, and N2 balance).

Expired gases for each exercise test were processed using five different sampling intervals: 15-s block averages (15sBl), 30-s block averages (30sBl), 15-breath block averages (15bBl), 15-breath moving averages (15bMov), and 30-s block averages time-aligned to the end of exercise (30sRetro). A block average was defined as the averaging of all single-breath V˙O2 measurements within a fixed period (e.g., 15 s) or fixed breath count (e.g., 15 breaths) (20) without overlapping measurements. The 15-breath moving average was defined as the averaging of 15 consecutive single-breath V˙O2 measurements (i.e., breaths 1 through 15), then removing the first breath and adding the next breath to obtain a new average (i.e., breaths 2–16), repeating through the end of the test. The 15sBl and the 30sBl were processed directly from the metabolic cart. The 15bBl, the 15bMov, and the 30sRetro were processed using the breath-by-breath data from the metabolic cart and Excel (Microsoft Corporation, Redmond, WA). These gas sampling intervals were chosen based on recommendations and frequency of use within the literature (2,8,10,13,14,20,21). The highest single-recorded V˙O2 value for each gas sampling interval was deemed V˙O2peak and used for subsequent analysis. The term V˙O2peak, as opposed to V˙O2max, was used to describe the highest V˙O2 for each sampling interval because it is not yet known which sampling intervals underestimates, correctly measures, or overestimates the true maximal oxygen consumption rate of the exercising individual.

Statistical Analysis

All statistical analyses were completed using SPSS version 22 (SPSS, Chicago, IL). For V˙O2peak comparisons within protocols for both subject samples, one-factor (gas sampling interval) repeated-measures ANOVAs were used. The Huynh–Feldt procedure was used when the assumption of sphericity was violated (Mauchly’s test). The Bonferroni post hoc test was used to identify significant differences.

To analyze whether differences in V˙O2peak for specific gas sampling intervals were of similar magnitude between subject samples, percent difference scores were calculated for the significantly different gas sampling interval combinations from the Bruce protocol. Difference scores were analyzed using independent samples t-tests. Significance for all tests was set a priori at P < 0.05.


Difference in V˙O2peak with Gas Sampling Interval

First subject sample

The Bruce protocol showed a significant main effect because of sampling intervals on V˙O2peak (F = 10.981, P < 0.001, partial η2 = 0.478, power = 1.000). Mean and SD are presented in Table 1. Post hoc tests revealed that 15bMov was the highest V˙O2peak and significantly higher than both 30sBl (P = 0.002) and 30sRetro (P = 0.002). The 15bBl was also significantly higher than 30sRetro (P < 0.05).

First subject set.

For the SPV protocol, there was a significant main effect because of sampling interval on V˙O2peak (F = 30.421, P < 0.001, partial η2 = 0.717. power = 1.000). Mean and SD are presented in Table 1. Post hoc tests revealed that 15bMov was again the highest V˙O2peak and significantly higher than all other sampling intervals (15bBl, P = 0.001; 15sBl, P = 0.001; 30sRetro, P < 0.001; 30sBl, P < 0.001). The 15bBl was significantly higher than 30sRetro (P = 0.022) and 30sBl (P = 0.007). The 15sBl was significantly higher than 30sBl (P = 0.041). The 30sRetro was significantly higher than 30sBl (P = 0.029).

Second subject sample

The Bruce protocol showed a significant difference between sampling intervals in V˙O2peak (F = 4.142, P = 0.04, partial η2 = 0.408, power = 0.630). Post hoc tests revealed that 15bMov was the highest V˙O2peak and significantly higher than 30sBl (P = 0.013). The modified Astrand protocol showed a significant main effect because of the sampling interval on V˙O2peak (F = 4.464, P = 0.034, partial η2 = 0.427, power = 0.659). Post hoc tests revealed that 15bMov was the highest V˙O2peak and significantly higher than 15bBl (P = 0.039), 30sBl (P = 0.032), and 30sRetro (P = 0.002). Mean and SD for each sampling interval for both protocols are presented in Table 2.

Second subject set.

Protocol Comparison

There was a noticeably similar pattern of change in V˙O2peak among sampling intervals. In each protocol, 15bMov gave the highest V˙O2peak. Figure 1 shows the V˙O2peak of each sampling interval for Bruce and SPV protocols from the first subject sample (solid lines). There was a change in the second highest V˙O2peak reported from 15bBl in SPV to 15sBl in Bruce. The 30sBl and the 30sRetro consistently reported the lowest V˙O2peak values, yet these intervals alternated rank between protocols. Figure 1 also shows the V˙O2peak of each sampling interval for Bruce and modified Astrand protocols from the second subject sample (dotted lines). In each protocol, the pattern of V˙O2peak change between sampling intervals was the same with 15bMov giving the highest V˙O2peak and 30sBl giving the lowest.

V˙O2 by sampling interval separated by subject sample and protocol. Solid line with solid squares, first subject sample Bruce protocol. Solid line with solid diamonds, first subject sample SPV protocol. Dotted line with open squares, second subject sample Bruce protocol. Dotted line with open circles, second subject sample modified Astrand protocol.

Subject Sample Comparison

After identifying the sampling intervals that presented a statistically significant difference in V˙O2peak during the Bruce protocol, the percent difference (%diff) in V˙O2peak for these intervals (15bMov–30sRetro, 15bBl–30sRetro, and 15bMov–30sBl) was compared between subject samples. There was no significant difference between subject samples for %diff in 15bMov–30sRetro (P = 0.967, d = 0.048) and 15bBl–30sRetro (P = 0.356, d = 0.471). There was a significant difference between subject samples in %diff for 15bMov–30sBl (P = 0.047, d = 0.971). These values are presented in Table 3.

Percent difference scores for select sampling intervals between subject samples in Bruce protocol.


It was the objective of this study to identify any potential differences in V˙O2peak during an SPV protocol when altering the gas sampling interval. In agreement with our hypothesis, there was a significant difference in V˙O2peak between sampling intervals during the SPV protocol. These differences are similar to other studies in magnitude and in regard to which gas sampling interval presented the highest and lowest V˙O2peak values (1,16).

New to this study is that fact that altering the gas sampling interval in the same subject sample when they underwent a different CPX protocol (Bruce) did little to alter the sampling interval effect on V˙O2peak. Figure 1 shows a pattern of change in V˙O2peak between the two protocols that is very similar. This supports our hypothesis that the gas sampling interval effect on V˙O2peak is protocol independent. This is the first clear evidence that protocol choice may not complicate the issue of choosing an appropriate sampling interval. One other study analyzed gas sampling intervals in more than one protocol; however, no comparisons were made between protocols to identify any potential protocol effects (3).

It should be noted that the CPX protocols in this study all used the same mode of exercise, the treadmill. It is known that altering the mode of exercise can affect V˙O2peak, particularly when the mode involves a smaller muscle mass and localized muscular fatigue such as cycling (23). Nevertheless, it may be reasonable to speculate that the gas sampling interval effect on V˙O2peak would follow a similar pattern in a mode of exercise such as cycling. Studies would need to be done to verify this assertion. At this time, one should be careful in generalizing the findings of this study to modes of exercise other than the treadmill. Also, the protocols chosen in this study have previously been used in our laboratories (8,21) and others (17) to assess V˙O2peak in healthy subjects with success. CPX protocols that are not appropriate for a given subject population because of variables such as large incremental work rate change between stages and test duration outside of what is recommended may not follow the same effects of gas sampling interval that occurred in the current study.

In an effort to verify the lack of difference that the CPX protocol has on the gas sampling interval, an independent subject sample undergoing a Bruce and modified Astrand protocol was analyzed. Our hypothesis was supported; the magnitude of change in V˙O2peak with alteration in gas sampling interval was similar in the second subject sample to that of the first sample. Figure 1 shows that the second subject sample had a similar gas sampling effect in both protocols and the profile of the change was very similar to the one seen for the first subject sample. The gas sampling intervals with significant protocol differences within a subject sample were compared between subject samples in terms of their magnitude of change. Only the 15bBl–30sRetro differences were not of the same magnitude between subject samples. This further substantiates the idea that the CPX protocol has limited influence on the gas sampling interval effect. The consistency in results between the two independent samples also provides evidence for the generalizability of the results among healthy, moderately fit individuals. Other studies looking at the gas sampling interval effect on V˙O2peak for specific populations have shown similar trends (higher V˙O2 with shorter intervals and lower with longer intervals) as was seen in the current study (11,23). One study that did not follow a similar pattern of change in V˙O2peak as the present study was completed in metabolic syndrome patients. Thomson et al. (24) reported similar V˙O2peak in 15bMov, 15-s retrograde, and 30sRetro, with only a slightly lower 60-s retrograde block averages. However, it should be noted that artifact such as swallowing, talking, and half-breaths were identified and removed before analyzing V˙O2peak. This reduction in variability may have reduced the potential for differences otherwise seen in the sampling intervals.

There are several aspects of CPX research that require attention in light of the current findings. First, research that includes V˙O2peak testing and reporting has, at times, failed to report the gas sampling interval used (12,25,26). The results of the current study support and highlight the continued need to report the gas sampling interval used when measuring V˙O2peak because of the differences that exist between intervals.

Second, the intrastudy differences in V˙O2peak may be practically significant in certain populations. Some may argue that the differences in V˙O2peak in the current study are not of practical significance as they rarely exceeded the expected measurement error of approximately 1.1–2.2 mL·kg−1·min−1 (2%–4%) (9,19), the largest difference was between 15bMov and 30sBl in the modified Astrand of the second subject sample reaching 3.3 mL·kg−1·min−1. We would like to point readers to the study by Smart et al. (23) who saw a smaller absolute change than the current study. This difference with alteration in sampling interval was of practical importance because the subject population was heart failure patients who had V˙O2peak values close to the heart transplant criteria cutoff. This supports the idea that changes in V˙O2peak with alterations in sampling interval are practically worthwhile, perhaps in some populations more than others.

Third, when looking at interstudy comparisons, researchers must be cautious when making comparisons to studies that used different sampling intervals or did not report a sampling interval. This would be particularly important for review articles. With the present ideas in mind, one could justifiably argue that serendipitously comparing “V˙O2max” or “V˙O2peak” interstudy, or against population reference values, has created miscalculation and misreporting in literature. Gas sampling intervals are not reported when choosing population-specific reference values; the actual study in which the references were synthesized would need to be accessed for clarification. It is equally as important to realize the potential for a confounding effect of different gas sampling intervals within a multisite study.

Fourth, most studies researching gas sampling intervals have focused on the incidence of a V˙O2 plateau, yet other measures are affected by gas sampling intervals. Measures of anaerobic threshold and RER that rely on these gas sampling measurements may also be altered. RER has been shown to change in heart failure patients with the manipulation of the gas sampling interval, with practical significance (23).

Lastly, these shortcomings in V˙O2max measurement, in part, highlight the need for maximal values to be reported as V˙O2peak, reserving the term V˙O2max for the theoretical maximal. We suggest this be the case until it is determined which gas sampling interval most appropriately represents true maximal oxygen consumption within the exercising individual. In addition, we are not recommending which specific gas sampling interval may be most appropriate for SPV or more traditional, staged maximal testing strategies presently. We can only say that changes in V˙O2peak between sampling intervals appear constant across protocols and younger, healthy subjects.

Limitations to the current study include a single exercise modality, as discussed earlier, and the coefficient of variation (CV) for the laboratories from which the data are drawn. The day-to-day variability in V˙O2max, in part, lies with the CPX testing procedures used in a particular laboratory. This CV can be determined for a laboratory's procedures with a given subject sample and CPX testing protocol and can aid in the interpretation of changes in V˙O2max that may occur. The CV values for the laboratories in this study, for the specific subject sample and protocol combinations, are not known with certainty. Although steps are taken to follow well-documented procedures, to use a fit and familiarized sample, and to be transparent in methods, these CVs may vary from the expected 2%–4% measurement error (9,19) because of unstandardized procedures that still exist, i.e., warm-up protocols, work rate increments, and methods for choosing the value to represent V˙O2max. When possible, the meaningfulness of the change in V˙O2max between CPX tests should consider the laboratory's CV in addition to the statistical significance and effect size.


It is evident that sampling intervals can significantly affect the V˙O2peak value reported during CPX testing. It appears that the CPX protocol used to elicit V˙O2peak has little, if any, effect on the manner by which gas sampling interval alters V˙O2peak. On the basis of the findings of this study, there is a need for continued research that focuses on gas sampling interval. Also, research is needed to further substantiate the idea that the gas sampling interval effect on V˙O2peak is largely independent of mode-specific protocols and population. Research needs to examine the degree to which the gas sampling intervals can alter the level of significance of a treatment effect on V˙O2peak. Most importantly, the greatest theoretically sound gas sampling interval(s) for measuring oxygen consumption under a given situation should be identified, and subsequent standardization of CPX testing using this empirically based interval should serve as a guide for future literature so that values can be reported as V˙O2max with increased certainty and compared within and between studies with reduced error. Finally, we propose that future literature should always include sampling interval used when reporting metabolic data so that comparison among research can align appropriately, even if we are not able to agreeably settle upon a consensus on this matter.

There are no funding sources to disclose. The authors have no conflicts of interest to declare. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The authors thank all those who contributed to the original studies. The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Astorino TA. Alterations in V˙Omax and the VO plateau with manipulation of sampling interval. Clin Physiol Funct Imaging. 2009;29(1):60–7.
2. Astorino TA, McMillan DW, Edmunds RM, Sanchez E. Increased cardiac output elicits higher V˙O2max in response to self-paced exercise. Appl Physiol Nutr Metab. 2015;40(3):223–9.
3. Astorino TA, Robergs RA, Ghiasvand F, Marks D, Burns S. Incidence of the oxygen Plateau at V˙O2max during exercise testing to volitional fatigue. J Exerc Physiol Online. 2000;3(4):1–12.
4. Balady GJ, Arena R, Sietsema K, et al. Clinician's guide to cardiopulmonary exercise testing in adults: a scientific statement from the American Heart Association. Circulation. 2010;122(2):191–225.
5. Borg G. Borg's Perceived Exertion and Pain Scales. Champaign (IL): Human Kinetics; 1998.
6. Bruce RA. Exercise testing of patients with coronary heart disease. Principles and normal standards for evaluation. Ann Clin Res. 1971;3(6):323–32.
7. Flowers TG, Garver MJ, Scheadler CM, et al. The impact of simulated altitude on selected elements of running performance. In: Proceedings of the International Journal of Exercise Science: Conference Proceedings. 2015. pp. 36.
8. Hanson NJ, Scheadler CM, Lee TL, Neuenfeldt NC, Michael TJ, Miller MG. Modality determines V˙O2max achieved in self-paced exercise tests: validation with the Bruce protocol. Eur J Appl Physiol. 2016;116(7):1313–9.
9. Howley ET, Bassett DR Jr, Welch HG. Criteria for maximal oxygen uptake: review and commentary. Med Sci Sports Exerc. 1995;27(9):1292–301.
10. Hunt KJ, Anandakumaran P, Loretz JA, Saengsuwan J. A new method for self-paced peak performance testing on a treadmill. Clin Physiol Funct Imaging. 2016; doi:10.1111/cpf.12390.
11. Johnson JS, Carlson JJ, VanderLaan RL, Langholz DE. Effects of sampling interval on peak oxygen consumption in patients evaluated for heart transplantation. Chest. 1998;113(3):816–9.
12. Kelly M, Gastin PB, Dwyer DB, Sostaric S, Snow RJ. Short duration heat acclimation in Australian football players. J Sports Sci Med. 2016;15(1):118–25.
13. Mauger AR. V˙O2max is altered by self-pacing during incremental exercise. Eur J Appl Physiol. 2013;113(2):541–2.
14. Mauger AR, Sculthorpe N. A new V˙O2max protocol allowing self-pacing in maximal incremental exercise. Br J Sports Med. 2012;46(1):59–63.
15. McArdle WD, Katch FI, Pechar GS. Comparison of continuous and discontinuous treadmill and bicycle tests for max V˙O2. Med Sci Sports. 1973;5(3):156–60.
16. Midgley AW, McNaughton LR, Carroll S. Effect of the V˙O2 time-averaging interval on the reproducibility of V˙O2max in healthy athletic subjects. Clin Physiol Funct Imaging. 2007;27(2):122–5.
17. Miller GS, Dougherty PJ, Green JS, Crouse SF. Comparison of cardiorespiratory responses of moderately trained men and women using two different treadmill protocols. J Strength Cond Res. 2007;21(4):1067–71.
18. Myers J, Walsh D, Sullivan M, Froelicher V. Effect of sampling on variability and plateau in oxygen uptake. J Appl Physiol (1985). 1990;68(1):404–10.
19. Robergs R, Burnett A. Methods used to process data from indirect calorimetry and their application to V˙O2max. J Exerc Physiol Online. 2003;6(2):44–57.
20. Robergs RA, Dwyer D, Astorino T. Recommendations for improved data processing from expired gas analysis indirect calorimetry. Sports Med. 2010;40(2):95–111.
21. Scheadler CM, Devor ST. V˙O2max measured with a self-selected work rate protocol on an automated treadmill. Med Sci Sports Exerc. 2015;47(10):2158–65.
22. Shephard RJ, Allen C, Benade AJ, et al. The maximum oxygen intake. An international reference standard of cardiorespiratory fitness. Bull World Health Organ. 1968;38(5):757–64.
23. Smart NA, Jeffriess L, Giallauria F, et al. Effect of duration of data averaging interval on reported peak V˙O2 in patients with heart failure. Int J Cardiol. 2015;182:530–3.
24. Thomson AC, Ramos JS, Fassett RG, Coombes JS, Dalleck LC. Optimal criteria and sampling interval to detect a V˙O2 plateau at V˙O2max in patients with metabolic syndrome. Res Sports Med. 2015;23(4):337–50.
25. Tota Ł, Maciejczyk M, Pokora I, Cempla J, Pilch W, Pałka T. Changes in endurance performance in young athletes during two training seasons. J Hum Kinet. 2015;49:149–58.
26. Valenti G, Bonomi AG, Westerterp KR. Multicomponent fitness training improves walking economy in older adults. Med Sci Sports Exerc. 2016;48(7):1365–70.


© 2017 American College of Sports Medicine