Critical power (CP) is a mechanical measure corresponding to the maximal steady state for blood lactate or oxygen uptake (V˙O2) (6). The traditional method of deriving CP, as originated by Monod and Scherrer (24), was determined by a minimum of four exhaustive square-wave exercise bouts to establish the hyperbolic power–time (duration) relationship (12). Where anaerobic work capacity (W′) defines the curvature constant, the asymptote of the power–duration relationship identifies the CP (24,25). Poole et al. (27) observed that square-wave exercise greater than CP yet below the power evoking V˙O2max (W peak) during a graded exercise test (GXT) still evoked V˙O2max (35).
Conventionally, lactate or gas exchange threshold (GET) defines the “heavy exercise” domain, characterized by a V˙O2 slow component and delayed steady-state lactate and V˙O2 kinetics (20). Conversely, CP demarcates the “severe exercise” domain characterized by a V˙O2 slow component that rises continually toward V˙O2max (27). Hill et al. (13) introduced a fourth domain or ceiling of the severe domain at ∼135% W peak. When their subjects engaged in square-wave exercise >135% W peak, exhaustion occurred before the achievement of V˙O2max; on the other hand, lower intensity bouts yielded times to V˙O2max conforming to the same power–duration relationship greater than CP.
Emergence of the verification bout after a GXT has refocused the practice of deriving “true” V˙O2max values (28). Traditional secondary criteria (16), most commonly the attainment of 10 bpm within the age-predicted HRmax and exceeding an RER of 1.10 (23), have been replaced by the attainment of a similar V˙O2 value in a different exhaustive exercise mode (e.g., difference of <3% for relative V˙O2 values) (21,23). Although it has been postulated that intensities greater than W peak are needed to establish true V˙O2max (15,26), loads greater than CP and less than W peak evoke a V˙O2 slow component to V˙O2max (27) and are suitable intensities for verifying V˙O2max (7,21,29). Because W peak precedes V˙O2max and end power in a GXT, assumed as a 1-min delay, verifications expressed as percentages greater than end-GXT power may encroach into the fourth domain (extreme) (6), especially with same-day verification bouts. The ∼135% W peak criterion, established as the ceiling of the severe domain (13), represents a mean intensity value for recreationally trained subjects (∼40 mL·kg−1·min−1) and a value derived from subjects not having completed a GXT preceding the exhaustive bout.
Although the square-wave protocol for verifying V˙O2max has been researched by numerous laboratories since its inception (22), less is known about the variability of V˙O2max measures attained from an all-out exercise bout. Burnley et al. (5) reported that a 3-min all-out exercise test (3 MT) yielded a valid measure of V˙O2max and an end-power congruent with CP or the maximal steady state for lactate and V˙O2. The 3 MT emerged from a 90-s all-out exercise test (8) that assessed W′ for 90 s reliably but that overestimated CP (9). Subsequently, a 3 MT that integrated power for 150 s, greater than CP, was introduced as a method to estimate W′ and CP along with V˙O2max, without the need for multiple exhaustive power–duration bouts (30).
Resistance for the original 3 MT was derived by determining mean power evoking GET (W GET) and W peak or 50% Δ, as established with a prior GXT. Thus, we presume assessing expired gases during the 3 MT would be for the purpose of verifying V˙O2max and not establishing V˙O2max initially; however, comparisons of variability with the GXT were not reported. The present study therefore assessed variability of V˙O2max from the 3 MT with the results of a GXT in comparison with V˙O2max evoked with an exhaustive square-wave bout. Second, because reliability data on CP and W′ were reported on the 90-s all-out exercise test (8) but not the 3 MT, we also evaluated the test–retest reliability of mechanical measures (i.e., work and power) derived using the 3 MT.
A sample of six males and five females (age = 22 ± 2 yr, height = 1.71 ± 0.07 m, body mass = 70.5 ± 10.8 kg) volunteered to take part in our study. Activity levels of the group consisted of runners (n = 5), cyclists (n = 2), and recreationally trained individuals (n = 4); however, each subject had prior experience exercising on a cycle ergometer. Health history screening forms were completed by each subject. The results from the forms were used to detect any preexisting medical conditions that would impose a serious risk to the subject or compromise results of the study (e.g., asthma, surgery of the lower extremity). Subjects verbally verified their compliance with the study’s guidelines, which included abstaining from the use of alcohol or engaging in heavy exercise 24 h preceding each visit. Subjects also indicated that they ate a similar diet, were hydrated properly, and avoided the use of caffeine 3 h before each visit; however, we did not monitor caloric or macronutrient intake via dietary recall. Subjects provided informed written consent to participate in the study, and all procedures were preapproved by our institutional review board.
Design and equipment
The subjects visited the laboratory on three separate occasions. Visit 1 was for the purpose of administering a GXT with an exhaustive square-wave verification bout. Visits 2 and 3 were for the purpose of conducting the 3 MTs. All tests were conducted using the same electronically braked cycle ergometer (Lode Excalibur Sport; Groningen, The Netherlands). The saddle and handlebar heights/positions were adjusted to each subject’s riding comfort and recorded for use throughout the study. Two subjects who were experienced cyclists opted to use their own cycling shoes and pedals, whereas the remaining subjects used pedal straps. Expired gas exchange data were collected on each visit using the same metabolic analyzer (ParvoMedics TrueOne; Sandy, UT) and evaluated using a 15-s sampling. Filter replacement and calibration were performed between tests according to the manufacturer’s guidelines.
Each subject completed a custom GXT predicted to evoke V˙O2max at 10 min and exhaustion shortly thereafter. Estimated V˙O2max was derived with a physical activity rating (range = 0–10) (11) and a regression equation (18). Subsequently, W peak was estimated using a metabolic equation (1) and divided by the 10-min duration of the GXT to yield a custom gradation. Gradations were between 17 and 33 W·min−1, and self-selected pace (desired cadence) was between 70 and 90 rpm. RPE, scaled 6 to 20 (4), was obtained before advancing stages, and strong verbal encouragement was provided as RPE exceeded 16 and beyond. When the desired cadence declined >10 rpm for 10 s, the GXT was terminated, and a setting of 50% end power was programmed manually on the ergometer for a 3-min cooldown phase. Upon completion of the cooldown, an intensity of two stages minus end power was set on the ergometer, and an exhaustive severe-intensity verification bout was performed (21). RPE was administered every 30 s, and strong verbal encouragement was provided. The verification bout terminated when the desired cadence declined >10 rpm for 10 s. We used a difference >3% relative V˙O2 (mL·kg−1·min−1) between the GXT and verification (21) as our criterion to retest a subject. Such retesting was necessary for one subject.
Using data from the GXT, GET was detected using the V-slope method (3). Mechanical measures of W GET and W peak were derived linearly as 1 min preceding the evoked V˙O2. The precision of power measures corresponding to GET/V˙O2max was 4.5–8.25 W (i.e., 25% of the gradation for the GXT), given that metabolic data were averaged to four samples per minute. For example, the interpolated power evoking a V˙O2 measure at 11.5 min for a 30-W·min−1 protocol would be 315 W. W GET, as a percentage of CP established for each trial of the 3 MT, also was calculated because previous investigators (10) have reported that W GET is ∼75% of CP in trained cyclists.
Three-minute all-out test
Resistance for the 3 MT was derived in accordance with the original investigators’ procedures (5,30). In brief, 50% Δ was divided by the desired cadence squared (rpm2) to derive a linear factor (∼3%–6% body mass (kg)). The product of the linear factor, gravitational force (nota bene (mark well), rounded to 10 m·s−2 by the ergometer’s software), and body mass was used to establish torque (N·m) for resistance of the 3 MT. We programmed a 50-W intensity for a 30-s lead-up, whereby during the last 5 s, the subjects were encouraged to build up their cadence to prohibit initiating the 3 MT from a dead start. As soon as the test commenced, the subjects were encouraged to pedal all-out with a maximal effort for 3 min. The test was terminated at 3 min 5 s to ensure a full 3-min sample was gathered. A programmed 3-min recovery at 75 W followed the 3 MT to prevent syncope.
Data from the 3 MT were exported from the cycle ergometer and retrieved in Microsoft® Excel® (Redmond, WA). Data were recorded at 6 Hz (default setting for sampling rate) and had visible noise that was more apparent in the latter portion of the 3 MT and in subjects who cycled less frequently as part of their regular exercise regimen. We attributed the noise to an uneven velocity of cycling strokes during the test (known colloquially as pedaling squares). Such noise was lower for subjects using clipped-in riding shoes as opposed to pedal straps. Prior investigators (8) using all-out exercise to estimate W′ have shown filtering in their methods; however, precise filtering calculations were not reported. We opted to apply a two-pass Butterworth-type filter using 5-Hz cutoff frequency, as described by Winter (36) (see representative subjects in Fig. 1). Presuming that some laboratories may use the raw data to administer the 3 MT, we report the reliability analyses of both our raw and filtered data.
The parameter of CP from the 3 MT was estimated using the mean power between 150 and 180 s (5,30). To estimate W′, we calculated the integral of work (J) above CP for the initial 150 s, as described by the original investigators. We also calculated mean power for the first 150 s and mean power greater than CP for the first 150 s. For each trial, power–time points were modeled, using t = 1[(power − CP)/W′], where t is time in seconds, power and CP are in watts, and W′ is in joules (33) (Fig. 2).
V˙O2max values for the GXT, verification bout, and first and second 3 MTs were analyzed using repeated-measures ANOVA. Paired-samples t-tests were used to assess W′ and 150-s mean power data between the 3 MTs. Reliability analyses were conducted using typical error (TE), coefficient of variance (CV) and intraclass correlation coefficients (α) (14). Summary statistics are reported as mean ± SD. Statistical significance was set at the 0.05 level.
W GET, V˙O2max, and data corresponding to 50% Δ from the GXT, along with work and power values derived from each 3 MT, are reported in Table 1. W GET from the GXT was 75% ± 14% and 73% ± 12% of CP (P = 0.55), as estimated with filtered data from trials 1 and 2 of the 3 MT, respectively. Power at 50% Δ was not different (F = 1.10, P = 0.37) from CP derived from either 3 MT.
The W′ parameter calculated using the filtered data from the first 3 MT was higher (t = 2.62, P = 0.02) than the second 3 MT (Table 1). Conversely, mean power for 150 s and mean power greater than CP for 150 s did not differ (P > 0.05) between trials (descriptive statistics also in Table 1). Test–retest reliability was stronger for the filtered versus the raw data derived from the 3 MT (Table 2). Power–duration data (Fig. 2) estimated from each 3 MT (filtered data) at 150, 300, 450, and 600 s did not differ (P > 0.05).
Relative V˙O2max (mL·kg−1·min−1) for the GXT (46.9 ± 10.8) and square-wave verification bout (47.3 ± 11.6) did not differ significantly (F = 2.83, P = 0.09) from the V˙O2max values derived from the first (47.3 ± 10.6) or second 3 MT (49.7 ± 10.9) trials. Figure 3 depicts V˙O2–time data for the GXT–verification bout and a 3 MT for a representative subject. Durations for GXT and square-wave bouts were 10.75 ± 1.20 and 3.43 ± 0.61 min, respectively. The variability for V˙O2max between the GXT and the square-wave bout (α = 0.99, TE = 1.16 mL·kg−1·min−1, CV = 2.8%) was lower in comparison with the variability analyses between the data derived from the GXT and the data derived from either 3 MT (trial 1: α = 0.97, TE = 2.03 mL·kg−1·min−1, CV = 5.5%; trial 2: α = 0.94, TE = 2.69 mL·kg−1·min−1, CV = 6.7%).
The primary aim of this study was to examine the reliability of the mechanical measures derived from the 3 MT using a Lode Excalibur cycle ergometer. Concerning estimates of CP, despite observing identical between-trial mean powers (Table 1), our absolute reliability (TE = 15.3 W, CV = 6.7%) was not as strong as the results reported by Burnley et al. (5) (TE = 7 W, CV = 3%), although TE values of ∼11 W have been reported by this group (31). From the present study, we observed TE was much greater if raw data were used (Table 2).
Concerning estimates of W′, data varied significantly between trials (Tables 1 and 2). In assessing the variability in units of power for 150 s (i.e., 11.1 W for total power), the absolute variability was comparable to that of CP (15.3 W). The CV, for either W′ or power greater than CP for 150 s, was considerable (i.e., ∼21% of unexplained variance for each metric). Conversely, CV for average power during 150 s was 4%, less than the 7% value observed for CP.
Although we observed significant group differences in W′ estimates yielded from the 3 MT, estimates of W′ have been close to approaching statistical significance in another study (i.e., P = 0.07) (32). Indeed, our findings indicated that W′ and power greater than CP for 150 s, from the 3 MT, are compounded by the unexplained variance of two metrics: power for 150 s and estimation of CP. These findings indicate that W′ per se lacks sensitivity. Conversely, average power during 150 s (Table 2) was extremely reliable and is recommended for rank ordering and monitoring training-induced power adaptations.
The recent training study by Vanhatalo et al. (31) provides proof, in principle, for the above-mentioned recommendations. The investigators examined the 3 MT and the power–time relationship modeled from three exhaustive square-wave bouts before and after a 4-wk interval training program. The 3 MT was sensitive to detect training-induced adaptations in total power, CP, and power durations, yet no training-induced adaptations in W′ were noted. Indeed, W′ measures estimated from the 3 MT increased by ∼600 J, whereas W′ estimated from the exhaustive square-wave bouts decreased by ∼1700 J! Although the authors attributed the disparity to the complexities of quantifying mechanisms mediating W′, we suggest that the disparity is related to unexplained variance in either method of determining W′. Exhaustive times for a single square-wave bout are a measure prone to high variability (34). By extension, multiple trials, each with inherent unexplained variance, suggest that W′ calculated using the traditional method is potentially less reliable than W′ assessed using the 3 MT. Despite the variability of the W′, when combined with CP, the ability to model the power–duration relationship consistently justifies continual assessment of W′ (Fig. 2).
The 3 MT was introduced as a technique to establish V˙O2max and the maximal steady state for V˙O2 (5). The procedures of that investigation involved determination of V˙O2max using a GXT where expired gases were averaged to 30 s. On the basis of a nonsignificant between-test difference, the investigators concluded that the 3 MT could attain V˙O2max. The extent to which the between-test measures varied with either absolute or relative statistics was not reported. Moreover, the investigators used a 30-W·min−1 protocol on a sample of subjects with absolute V˙O2max values measured as low as 2.52 and as high as 4.71 L·min−1. Such a range in absolute values assuredly altered the GXT duration.
Our study used a square-wave verification bout to determine true V˙O2max, as recommended by others (22,28). Thus, when an individual GXT differed more than 3% from the square-wave bout, we were able to determine the need to retest the GXT (as was our case for one subject). Such a procedure was not implemented by Burnley et al. (5) (NB, emergence of square-wave verification method (28) occurred near the time of their publication). We compared the V˙O2max values from the 3 MT with the GXT and determined that those comparisons were more variable than the GXT and square-wave protocol comparison (a greater variability of ∼1–2 mL·kg−1·min−1) (Fig. 3). We conclude that the all-out mode of exercise to verify V˙O2max is less consistent than conducting the square-wave verification bout.
It has been established previously that W GET occurs at a value of ∼75% of CP for cycle ergometry, when CP is established using a series of exhaustive power–duration bouts (19). Because the 3 MT provides an estimate of CP (5,30), it is reasonable that W GET, as estimated using the 3 MT, also would occur at ∼75% of estimated CP. Indeed, ∼75% CP estimated using the 3 MT was observed in fit subjects (10) and is consistent with our subjects of more considerable variability in aerobic fitness. The present 3 MT protocol, as derived by Burnley et al. (5), requires a preliminary GXT to establish 50% Δ; thus, both GET and W peak are known, and there is no need for an estimate.
Research on indirect methods of deriving percentage(s) of body mass for the 3 MT may expedite use of this innovative test. With the 30-s Wingate test, for instance, load-per-body-mass values of 7.5% and ∼10% have been recommended for sedentary and physically active/athletic individuals, respectively (17). Linear factors for the 3 MT in the present study were ∼3% body mass for lower fitness levels (V˙O2max < 40 mL·kg−1·min−1) and ∼4% to 5% body mass for higher fitness levels (50 to >60 mL·kg−1·min−1). We anticipate, on the basis of our findings, that the W′ estimate would vary regardless of two different loads used; however, CP estimates would be robust and similar. The primary concern of a load-per-body-mass recommendation for estimating CP with all-out exercise is that of end-exercise cadence. Barker et al. (2), for instance, reported that CP modeled with exhaustive bouts at 100 versus 60 rpm resulted in ∼18 W lower CP values. By extension, a 3 MT using too low of a load (relative to body mass) may end with higher values of revolutions per minute and yield low estimations of CP. The solution may be to derive a protocol with a feature that verifies a true 3 MT was performed (i.e., a confirmatory square-wave bout that ends within a time frame predicted by the 3 MT). The utility of the 3 MT is clear, and more research on refining the use of this test is merited.
In summary, our results indicate that the 3 MT yields reliable measures of CP and, by extension, reliable estimates of the power–duration relationship. Our data, along with others (31), indicate that total power for 150 s serves as a better metric of detecting changes in short-term power, a performance influenced both by CP and W′. Finally, we conclude the square-wave bout as opposed to 3 MT serves as a more suitable mode for verifying true V˙O2max.
This work has not been supported by any funding agency.
The authors report no conflict of interest.
Results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. Baltimore (MD): Lippincott, Williams & Wilkins; 2010. p. 158.
2. Barker T, Poole DC, Noble ML, Barstow TJ. Human critical power–oxygen uptake relationship at different pedalling frequencies. Exp Physiol. 2006; 91: 621–32.
3. Beaver WL, Wasserman K, Whipp BJ. A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol. 1986; 60: 2020–7.
4. Borg G. Perceived exertion as an indicator of somatic stress. Scand J Rehab Med. 1970; 2: 92–8.
5. Burnley M, Doust JH, Vanhatalo A. A 3-min all-out test to determine peak oxygen uptake and the maximal steady state. Med Sci Sports Exerc. 2006; 38 (11): 1995–2003.
6. Burnley M, Jones AM. Oxygen uptake kinetics as a determinant of sports performance. Eur J Sport Sci. 2007; 7: 63–79.
7. Day JR, Rossiter HB, Coats EM, Skasick A, Whipp BJ. The maximally attainable V˙O2
during exercise in humans: the peak vs. maximum issue. J Appl Physiol. 2003; 95: 1901–7.
8. Dekerle J, Brickley G, Hammond AJ, Pringle JS, Carter H. Validity of the two-parameter model in estimating the anaerobic work capacity. Eur J Appl Physiol. 2006; 96: 257–64.
9. Dekerle J, Vanhatalo A, Burnley M. Determination of critical power from a single test. Sci Sports. 2008; 23: 231–8.
10. Francis JT Jr, Quinn TJ, Amann M, LaRoche DP. Defining intensity domains from the end power of a 3-min all-out cycling test. Med Sci Sports Exerc. 2010; 42 (9): 1769–75.
11. George JD, Stone WJ, Burkett LN. Non-exercise V˙O2max
estimation for physically active college students. Med Sci Sports Exerc. 1997; 29 (3): 415–23.
12. Hill DW. The critical power concept: a review. Sports Med. 1993; 16: 237–54.
13. Hill DW, Poole DC, Smith JC. The relationship between power and the time to achieve V˙O2max
. Med Sci Sports Exerc. 2002; 34 (4): 709–14.
14. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med. 2000; 30: 1–15.
15. Howley ET. V˙O2max
and the plateau—needed or not? Med Sci Sports Exerc. 2007; 39 (1): 101–2.
16. Howley ET, Bassett DR Jr, Welch HG. Criteria for maximal oxygen uptake: review and commentary. Med Sci Sports Exerc. 1995; 27 (9): 1292–301.
17. Inbar O, Bar-Or O, Skinner JS. The Wingate Anaerobic Test. Champaign (IL): Human Kinetics; 1996. p. 16–22.
18. Jackson AS, Blair SN, Mahar MT, Wier LT, Ross RM, Stuteville JE. Prediction of functional aerobic capacity without exercise testing. Med Sci Sports Exerc. 1990; 22 (6): 863–70.
19. Jones AM, Poole DC. Introduction to oxygen uptake kinetics and historical development of the discipline. In: Jones AM, Poole DC, editors. Oxygen Uptake Kinetics in Sport, Exercise and Medicine. New York (NY): Taylor & Francis; 2005. p. 3–36.
20. Jones AM, Vanhatalo A, Burnley M, Morton RH, Poole DC. Critical power: implications for the determination of V˙O2max
and exercise tolerance. Med Sci Sports Exerc. 2010; 42 (10): 1876–90.
21. Kirkeberg JM, Dalleck LC, Kamphoff CS, Pettitt RW. Validity of 3 protocols for verifying V˙O2max
. Int J Sports Med. 2011; 32: 266–70.
22. Midgley AW, Carroll S. Emergence of the verification phase procedure for confirming “true” V˙O2max
. Scand J Med Sci Sports. 2009; 19: 313–22.
23. Midgley AW, McNaughton LR, Polman R, Marchant D. Criteria for determination of maximal oxygen uptake: a brief critique and recommendations for future research. Sports Med. 2007; 37: 1019–28.
24. Monod H, Scherrer J. The work capacity of a synergistic muscle group. Ergonomics. 1965; 8: 329–38.
25. Moritani T, Nagata A, deVries HA, Muro M. Critical power as a measure of physical work capacity and anaerobic threshold. Ergonomics. 1981; 24: 339–50.
26. Noakes TD. Maximal oxygen uptake as a parametric measure of cardiorespiratory capacity: comment. Med Sci Sports Exerc. 2008; 40 (3): 585.
27. Poole DC, Ward SA, Gardner GW, Whipp BJ. Metabolic and respiratory profile of the upper limit for prolonged exercise in man. Ergonomics. 1988; 31: 1265–79.
28. Poole DC, Wilkerson DP, Jones AM. Validity of criteria for establishing maximal O2
uptake during ramp exercise tests. Eur J Appl Physiol. 2008; 102: 403–10.
29. Rossiter HB, Kowalchuk JM, Whipp BJ. A test to establish maximum O2
uptake despite no plateau in the O2
uptake response to ramp incremental exercise. J Appl Physiol. 2006; 100: 764–70.
30. Vanhatalo A, Doust JH, Burnley M. Determination of critical power using a 3-min all-out cycling test. Med Sci Sports Exerc. 2007; 39 (3): 548–55.
31. Vanhatalo A, Doust JH, Burnley M. A 3-min all-out cycling test is sensitive to a change in critical power. Med Sci Sports Exerc. 2008; 40 (9): 1693–9.
32. Vanhatalo A, Doust JH, Burnley M. Robustness of a 3 min all-out cycling test to manipulations of power profile and cadence in humans. Exp Physiol. 2008; 93: 383–90.
33. Whipp BJ, Huntsman DJ, Stoner N, Lamarra N, Wasserman KA. A constant which determines the duration of tolerance to high-intensity work. Fed Proc. 1982; 41: 1591.
34. Whipp BJ, Ward SA. Quantifying intervention-related improvements in exercise tolerance. Eur Respir J. 2009; 33: 1254–60.
35. Whipp BJ, Wasserman K. Oxygen uptake kinetics for various intensities of constant-load work. J Appl Physiol. 1972; 33: 351–6.
36. Winter DA. Biomechanics and Motor Control of Human Movement. Hoboken (NJ): Wiley; 2004. p. 11–50.