You are on page 1of 26

“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC

International Journal of Sports Physiology and Performance


© 2016 Human Kinetics, Inc.

Note. This article will be published in a forthcoming issue of the


International Journal of Sports Physiology and Performance. The article
appears here in its accepted, peer-reviewed form, as it was provided by
the submitting author. It has not been copyedited, proofread, or
formatted by the publisher.

Section: Original Investigation

Article Title: 3-Min All-Out Test in Swimming

Authors: Ming-Chang Tsai1,2 and Scott G. Thomas1

Affiliations: 1Health and Performance Laboratory, Graduate Department of Exercise Sciences,


Faculty of Kinesiology and Physical Education, University of Toronto, Toronto, Ontario,
Canada. 2Canadian Sport Institute Pacific, Victoria, British Columbia, Canada.

Journal: International Journal of Sports Physiology and Performance

Acceptance Date: March 9, 2016

©2016 Human Kinetics, Inc.

DOI: http://dx.doi.org/10.1123/ijspp.2015-0479
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

RUNNING TITLE: 3-Min All-Out Test in Swimming

3-Min All-Out Test in Swimming

Original Investigation

Ming-Chang Tsai1,2 and Scott G. Thomas1


Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

1
Health and Performance Laboratory, Graduate Department of Exercise Sciences, Faculty of
Kinesiology and Physical Education, University of Toronto, Toronto, Ontario, M5S 2W6,
Canada
2
Canadian Sport Institute Pacific, Victoria, British Columbia, V9E 2C5, Canada

Address for correspondence: Ming-Chang Tsai, Health and Performance Laboratory, Graduate
Department of Exercise Sciences, University of Toronto, Toronto, Ontario, M5S 2W6, Canada

Telephone number: +14169780672

E-mail: ming.tsai@mail.utoronto.ca

Abstract word count: 244

Text-only word count: 3783

Number of figures and tables: 6


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Abstract

Purpose: To validate the 3-min all-out test (3MT) protocol against the traditional critical speed

model (CSM) in front-crawl swimming. Methods: Ten healthy swimmers or triathletes

(mean±SD: age 35.2±10.5yr, height 176.5±5.4cm, body mass 69.6±8.2kg) completed 5 tests

(3MT, 100m, 200m, 400m, 800m) over 2 weeks on separate days. Traditional Critical Speed

(CS) and Anaerobic Distance Capacity (D’) were determined for each of the 3 traditional CSM

(linear distance-time: LIN, linear speed-time-1: INV, nonlinear time-speed: NLIN) from the 4 set
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

distance time trials. For the 3MT, CS was determined as the mean speed during the final 30s of

the test and D’ was estimated as the power-time integral above the critical speed. Results: Our

results indicated no significant difference between the CS estimates determined from the

traditional CSM and 3MT except for the INV model (INV p=0.0311). Correlations between

traditional CSs and 3MT were high (r = 0.95, p < 0.01) However, D’ differed and post hoc

analysis indicated that D’ estimated from 3MT was significantly lower than LIN (p=0.0052) and

NLIN (p<0.0001). Correlations were weak (r < 0.55, p > 0.1). In addition, Bland-Altman plots

between the traditional CSMs and 3MT CS estimates showed scattered points above and below

the zero line suggesting that there is no consistent bias of one approach versus the other.

Conclusions: 3MT is a valid protocol for swimming to estimate CS. The demonstrated

concurrent validity of the 3MT may allow more widespread use of CSM to evaluate participants

and responses to training.

Key Words: All-out, critical speed, testing, monitoring, performance, modelling


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Introduction

The 3-min all out exercise test (3MT) has recently gained popularity as a testing protocol

that can estimate Critical Power Model (CPM) parameters accurately with a single test1 to

address the time-consuming protocol of CPM proposed by Moritani.2 Based on the assumption

that the anaerobic capacity of an individual can be fully depleted during an all-out test,1 the

power output at the end of the test should represent that power which is sustainable by the

aerobic system alone, and this is the definition of Critical Power (CP). The test must be long
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

enough to deplete anaerobic stores and require reliance on energy production from aerobic

metabolism.3 Dekerle showed that with all-out exercise of 90 s in duration, power output at the

end of exercise is still considerably higher than CP.4 however in a longer all-out test, power

output may continue to fall to an end-power (EP) that would equal that associated with the CP

intensity. A 3 min all-out test was proposed by Burnley showing that the EP is a good

representation of the boundary between the heavy and severe exercise domains.5 Vanhatalo and

colleagues observed in their sample population that EP was not significantly different from, and

highly correlated with CP (p=0.37; r=0.99). The standard error for the estimation of CP using

EP was approximately 6W. Work completed above EP derived from the 3MT (15 kJ) was not

significantly different from W’ (16 kJ, p =0.35), and correlated with W’, (r=0.84).1

The CPM has been extended to activities such as running and swimming where power

output is difficult or impossible to measure. Speed may serve as the basis for an analogous

model, Critical Speed Model (CSM), applied to swimming in which critical speed (CS)

corresponds to CP and the fixed work capacity above CS (D’) corresponds to W’.6 As with other

sports such as cycling,7 running,6 rowing,8 and racket sports.9 The CSM model has been applied

to assess performance and fitness status in swimming.10, 11


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Similarly, 3MT has been applied to several endurance sports such as cycling,1 running,12

and rowing.13 However, the protocol has only been reported once to evaluate tethered swimming

which the authors acknowledged could alter stroke mechanics14 resulting in inaccuracy in

estimating the CSM parameters. Therefore, the objective of this study is to examine the

concurrent validity of the 3MT protocol in estimating CSM parameters. We hypothesized that

the CS and D’ derived from the 3MT will not differ from the traditional CSM estimates.
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

Methods

Participants

Ten healthy average trained club swimmers/triathletes (4 males and 6 females, mean±SD:

age 35.2±10.5yrs, height 176.5±5.4cm, body mass 69.6±8.2kg) volunteered to take part in this

study. The subjects had at least 2 years of experience competing in swimming and triathlon

races and trained from 5-10 h per week during the two months preceding the tests. Total weekly

swimming volume ranges from 6-12km and the 400 m time-trial time was 422.0±65.2 s.

The subjects were requested to refrain from participating in strenuous physical activity in

the 24 h prior to the test and were to refrain from consuming caffeine and alcohol 3 h before

reporting to the laboratory. They were informed about the study aims, the procedures and risks

associated with the tests, and all the participants gave their written and informed consent. The

study was approved by the University of Toronto Review Ethics Board and was conducted in

accordance with the Declaration of Helsinki.

Protocols

The traditional method of determining critical speed involves repeated exhaustive testing

efforts at different distances done over several days. The subjects were athletes who had

publically available results for competition (in 50m pool) of different distances from within the
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

past 2 weeks. One of the advantages of using competition results is its ease of data collection

which has been shown to be a viable protocol.15 By using these results we minimized the number

of tests the athlete had to complete, so the study protocol combined competition results and one

to three visits to the 50m pool for data collection totalling up to five tests (3MT, 100m, 200m,

400m, 800m) for each individual. The difference between dive and push-off start was shown to

be ~0.87 s16 which after adjustment resulted in no change in CS and minimal differences of 2-4%

(0.22 to 1.2 m) change in D’ compared to the unadjusted combined competition and time trial
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

results. Therefore the data were left unadjusted.

For the 3-min all-out test, the trial started with one min of standing rest in the pool before

the start of the 3-min all-out test. Verbal encouragement was given throughout the tests,

although no elapsed time feedback were given to the subjects during the test to avoid pacing.

Subjects were instructed and strongly encouraged to provide maximal effort at all times

throughout the test. Time splits were measured at every 25m to the nearest tenth of a second and

the test was stopped when the last time split recorded was more than 3 min and 5 s to ensure

gathering a full 3 min of data. An estimate of the subject’s CS and work done above CS was

determined using the three-minute protocol of Vanhatalo and colleagues1. For the time trial test,

each subject completed a series of one to two set-distance swims (100m, 200m, 400m, 800m)

that were missing from the competition results in with the goal set to achieve the fastest time

possible. Total time to complete the swim was recorded to the nearest tenth of a second by two

qualified timekeepers. Wakayoshi11 suggested that swimmers should complete distances of 50,

100, 200, and 400m with a maximum intensity. At this effort, swimmer should be able to

exhaust anaerobic energy stores at the end of the test, however, the energy cost increases

exponentially at speeds leading to an attainment of VO2max.17 To adjust for this increased cost,
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

the most appropriate duration of distance is between 3 to 15min when the 2-parameter model is

used while inclusion of a shorter duration distance of 1min could be better when a 3-parameter

model is used for CS calculation.18 We have included a rationale for choice of the swim lengths

(We selected these distances to avoid marked efficiency reductions observed with short duration,

high speed swims (< 1 minute; 50 meters)19 and to elicit test durations between 1 and 15 minutes

which has been shown to be a good approach to determine aerobic ability.20

All the tests were performed in front crawl with a push-off start in randomized order.
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

Data were reanalyzed accounting for the wall push (0.97s slower) and coffee performance effort

(~2% improvement), the change in CS estimates were between 0.001-0.006 m/s (0.1-0.6%)

which is within the CV of the swimming performance. (0.8% of CV for elite swimming

performance21) and considering the level of the subjects in our study, CV would be higher).

Time trial testing occurred at the same time of the day (±1 hr) for each swimmer in order to

minimize the effect of circadian variation on performance.22 There were 5 subjects had

competition results in the 100/200/400m events (2 in 100m, 1 in 100/400m, 1 in 100/200/400m

and 1 in 200/400m). Both 100m and 400m were in the morning within ± 1hr of the time trial

time. 2 of the subjects competed in the 200m which was done in early afternoon. There is a 7%

increase in swim bench power (~2% in speed) between 10am and 2pm23. In addition, the effect

of caffeine on 1500m swim is 1.79%24 and on 50yd swim is 0.87%25. Adjusting for these effects

and wall push start, the change in CS was between 0.0004 to 0.005 m/s (0.05% to 0.59%) which

is within the individual performance CV. Subjects had a minimum of 24 hours of rest between

tests and all testing and races were completed within 2 weeks. Before each test, subjects

performed their regular race warm up protocol lasting between 10-20 min and then 5-10 min of

rest.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Data Analysis

Three traditional critical speed models and a 3-min all-out test were utilized to derive

four different estimates of CS and D’ (Figure 1), where t is the time to exhaustion; S is the speed;

D is the total distance; CS is the critical speed, and D’ is the anaerobic distance capacity.

The linear total distance (LIN) model was based on the linear regression of total distance

versus time,26

𝐷 = 𝐷′ + 𝐶𝑆 ∙ 𝑡
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

Dividing the LIN model by t yields the linear speed model (INV) where a linear form

between S and t can be converted from the hyperbolic relationship,27

𝐷′
𝑆= + 𝐶𝑆
𝑡

A more intuitive and preferable version from an exercise performance point of view is the

nonlinear time to speed (NLIN) model (S >CS), in which time (t) is the obvious dependent

variable as it is dependent on the speed of the test and speed (S) is the explanatory variable,27

𝐷′
𝑡=
𝑆 − 𝐶𝑆

The 3-min all-out test (3MT) was proposed by Burnley on the basis of cycling data

showing that the end power is a good representation of the boundary between the heavy and

severe exercise intensity domain5 and it is highly correlated with critical power.1 We followed

the approach of Pettit12 who extended the test to running and showed that CS can be determined

as the mean speed during the final 30 s of the test, and the D’ is estimated as the speed-time

integral above the critical speed.


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Statistical Analysis

The CS and D’ estimates between all the models were compared using a one-way

repeated-measures analysis of variance mixed model in SAS (Version 9.4; SAS Institute, Cary,

NC) with a compounded symmetry assumption made on the variance-covariance structure. This

analysis approach was adopted based on the consistent variance between models. The Tukey

HSD post-hoc procedure was used to control for type I error in making multiple comparisons, to

determine the significant difference between the parameters estimated for all the model
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

parameters. Investigation of the residual plot showed a random scatter of points and normality

plot showed the residuals fall on a straight-line indicating the normality assumption was

appropriate for both CS and D’. The results from the Levene’s test showed no differences in

variance for CS and D’ (p=0.987 and p=0.262 respectively), therefore homoscedasticity

assumption was met. Mauchly’s Test of Sphericity indicated that the assumption of sphericity

has been violated (χ2=0.120, p=0.006). The Geisser and Greenhouse method was used to correct

for the violation (p=0.048). A sample size of 10 provided adequate power (=0.8) to detect a

difference of 0.03 m.s-1. Pearson correlation coefficients were calculated to assess the strength of

linear relationship between estimates. A Bland-Altman analysis was used to assess the level of

agreement between alternative estimates of CSM parameters. A range of agreement was defined

as mean bias ±2 SD with 100% percentage of values within the limits. The limits of agreement

of differences was defined a priori as 5% of CS which is clinically acceptable. One way

repeated-measures analysis of variance with Dunnett-Hsu post-hoc procedure was used to

compare parameters estimated at each of the time interval to the parameters estimated at 3 min,

as well as the comparing the parameters estimate from one time interval to the previous interval.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Results

The mean (±SD) times obtained during the 100, 200, 400, and 800 m swimming time

trials were 88.2 (±12.0), 192.1 (±28.6), 422 (±65.2), and 868.2 (±127.6) seconds. The boxplot

(Figure 2) shows the median CS and its variance with the three traditional CSMs and 3MT is

similar, however on the contrary, the median of all the D’ estimates differ and there is more

variance between models.

Table 1 provides a summary of CS and D’ estimates from the three traditional critical
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

speed model and 3-min all-out test. The repeated measures ANOVA indicated that there were

significant differences amongst the means of the 4 estimates of CS (F3,27 = 3.75, p=0.0227) and

the post hoc comparisons indicated that the INV model produced the highest estimate of CS

while 3MT produced the lowest estimate of CS resulting in a significant difference (p=0.0311)

between the two estimates. However, there were no other significant differences among the CS

estimates. Furthermore, repeated measures ANOVA indicated that there were significant mean

differences (F3,27 = 10.72, p<0.0001) among the 4 estimates of D’. Specifically, the post hoc

comparisons showed that the LIN (p=0.0052) and NLIN (p<0.0001) models produced

significantly higher estimates of D’ than did the 3MT.

The scatterplot showed a linear relationship between the 3MT and the traditional

parameter estimates and the correlations were statistically significant for CS (LIN r=0.956: p <

0.001, INV r=0.951: p < 0.001, NLIN r=0.959: p < 0.001) indicating that there was strong linear

relationship for CS. However, estimates of D’ from 3MT did not significantly correlate with the

D’ estimates calculated from the traditional CSMs (LIN r=0.336 p=0.342, INV r=0.515 p=0.127,

NLIN r=0.232 p=0.518) (Figure 3-5). Figures 3-5 also show the relationship and bias ±95%

limits of agreement between CSM parameter estimates calculated from 3MT and CSMs. The
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

standard error (SEE) between the CS3MT and CS estimates derived from LIN, INV, and NLIN

were ±0.107 m.s-1, ±0.110 m.s-1, and ±0.108 m.s-1 respectively. The p-value for Pitman’s test of

equality of variances were 0.777-0.991 for CS and 0.403-0.815 for D’ suggesting there is no

significant difference between variances for the 3MT and the traditional models’ parameter

estimates and thus no effect of 3MT parameter estimates beyond regression to the mean. Bland-

Altman plots (Figures 3-5) of CS between the traditional CSMs and 3MT indicate that the 95%

limits of agreement range from -0.11 to 0.08 m.s-1 (LIN: -0.09 to 0.07 m.s-1, INV: -0.11 to 0.05
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

m.s-1, NLIN: -0.08 to 0.08 m.s-1) and with all the points within the limits. The mean difference in

CS between 3MT and traditional CSMs are LIN: 1.19%, INV: 3.52%, NLIN: 0.2% of 3MT CS.

This range is within the value of 5% 3MT CS defined a priori as acceptable. The result did not

show the same for D’, LIN: 65.95%, INV: 31.35%, NLIN: 95.18%. The differences in variances

are LIN: 0.0016, INV: 0.0017, NLIN: 0.0017. The plots showed scattered points above and

below the zero line suggesting that there is no systematic bias of one approach versus the other.

However, the plots showed a consistent lower D’3MT value compared to D’ estimates calculated

from CSMs (Figures 3-5). CS values derived from the 3MT ranged from 0.71 to 1.10 m.s-1 and

D’ from 7.54 to 32.33 m.

The peak speed in the 3MT, typically attained within 10-15 s of the start of the test, was

1.36 ± 0.18 m.s-1. Speed was calculated at 15s interval by assuming a linear regression within

the 5m distance intervals (i.e. 20m to 25m) to the times that cross each of the 15s intervals (i.e.

14.10s and 18.20s). The speed at the 15s interval can be calculated by substituting the interval

time (i.e. 15s). The group mean speeds averaged every 15 s interval were compared and the

analysis indicated that there were significant differences amongst the means of the 12 speed

intervals (F11,141 = 32.28, p<0.0001). All the time intervals differed significantly from the last 15
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

s period (165-180 s), with the exception of the final 90 s, where the changes in speed were

0.0714 m.s-1 (95% confidence limits -0.0255, 0.1683 m.s-1), -0.0023 m.s-1 (95% CL -0.074,

0.1199 m.s-1), -0.0270 m.s-1 (95% CL -0.0699, 0.1239 m.s-1), 0.0550 m.s-1 (95% CL -0.0419,

0.1519 m.s-1), and 0.0087 (95% CL -0.0882, 0.1057 m.s-1) for the last 6 comparisons (Figure 6).

Thus the speed had stabilized in the last 90 s of the test, justifying the 30 s average being used as

CS. The stabilization of speed in the last 90 s of the 3MT suggests an all-out swimming test with

duration of 135 s is sufficient to estimate CS accurately. As for the worthwhile change


Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

compared to the last 30s stability of speed, we interpreted it in 2 ways; 1) at which time interval

will the CS be not be significantly different from the CS3MT, 2) the worthwhile change from one

interval compared to the interval before. Figure 6 compared the CS determined from each

interval to CP3MT, the worthwhile change was 0.11 m/s at 90s interval which showed a

significance difference from CP3MT. The CS differences were 0.0087-0.07 m/s for the intervals

from 90s to 180s. However, comparisons of CS from one interval to the previous interval

showed a change of 0.12 m/s at 60s interval when compared to 45s interval, all the intervals after

that interval did not show a significance change with the differences of 0.0087 m/s to 0.056 m/s.

Discussion

The results with our diverse sample of swimmers and triathletes demonstrate that a 3-min

all-out swimming test provides accurate estimates of traditional critical speed with correlation

coefficients greater than r = 0.95 and SEE of approximately 0.10 m.s-1. However, D’ estimates

calculated from 3MT were more variable and 50-75% lower than those estimated from the

traditional CSMs. These findings provide a foundation for the use of 3MT derived CS as a tool

for characterizing athletes’ aerobic fitness.


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

One advantage of the 3MT to the traditional CSM protocol is its ability to determine CS

in a time-efficient manner (single test vs. multiple tests). The standard error for the estimation of

critical power in cycling using end power was approximately 6-11 W or 2-5% of the mean end

power value1, 28
while using end speed calculated from 3MT in running estimated racing

performance within 2% for 1600 and 1500m.12 In our study, the standard error between CS3MT

and CS estimates was approximately 0.11 m.s-1 or 12% of the mean CS3MT value. The

coefficients of determination for the 3 traditional models used to estimate CS ranged from 0.891-
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

1.00. These findings were consistent with those reported (r2 = 0.818-1.00) in previous studies26,
29-31
irrespective of the models and indicating close relationship between speed and time.

While the estimate of W’ calculated from the 3MT in cycling was shown to be no

different from the estimate determined from the traditional model,1 the reliability of W’ estimates

is often observed to be lower than those for CP (coefficient of variance (CV) of 10.2% for CP

and 27.5% for W’)32 on fixed resistance, while de Lucas observed CVs of 5.7% and 7.9% for CP

and W’ respectively on isokinetic cycling mode.33 Our study showed a similar CV range of 14%

for CS and 24-31% for D’. We observed a significant difference between the estimates of D’

determined from LIN (p=0.0052) and NLIN (p<0.0001) and failed to confirm any relationship

(LIN r=0.336 p=0.342, INV r=0.515 p=0.127, NLIN r=0.232 p=0.518).

The D’3MT was lower than the 3 traditional CSM estimates (Table 1). A plausible

explanation could be the exponential increase in energetic cost with speed due to hydrodynamic

drag swimmers encounter leading to distortion of models.17 The distortion is evident in the 15 s

mean speed plot (Figure 6) where there is a short but drastic initial decrease in speed compared

to other reported studies. Previous studies comparing one time interval to the next found a

significant decrease to 90 s in rowing13 and 135 s in cycling.1 Our study showed a significant
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

decrease to 60 s before speed gradually settled in the last 120 s (Figure 6). The quicker decrease

in speed and shorter time in reaching asymptotic speed led to a smaller D’ (speed-time integral

bounded by CS) as shown in Table 1.

The difference in D’ estimates could also be explained partially by the inaccuracy in the

constant distance-time protocol of the traditional CSM in capturing the total distance completed.

The subjects may be unable to maintain the required speed at the end of the exhaustive tests,

however, if the exercise was to continue after the termination of the constant speed test, speed
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

would gradually decrease but remain above CS for some time (exercise can be continued if
8, 27
speeds are lowered) indicating that the total anaerobic metabolic capacity has not been

exhausted. This additional distance expenditure is not accounted for in the traditional CSM

which results in underestimating D’ and hence increases the disagreement between D’ estimates

from 3MT and LIN. Lastly, the work efficiency may not be constant within the range of test

intensities,19, 34 work efficiency would decrease as the time to exhaustion increases which would

further suggest an underestimate of the actual metabolic cost required to produce the additional

distance output.4

The wide range of fitness levels amongst our subjects is a strength for generalizability but

may also be a limitation. It has been shown that pain tolerance increases with aerobic fitness35

and consequent differences in swim speed may influence estimation of the CP3MT in the last 60-

90 s of the all-out test. Two possible scenarios could low pain tolerance reducing speed close to

the end of the test leading to leading to a lower CS or an early pain induced speed reduction

followed by an end sprint which raises CS. We do not have an independent measure of pain

tolerance to identify this possible effect. The effect of fatigue at the end of the test could

potentially alter the swimming stroke resulting in loss of propulsion efficiency. This could
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

possibly show up on the speed profile with another drop of end speed, however, we did not

observe such a trend in our data and in addition to our 3MT CS was 1-3% from the traditional

model CSs supporting 3MT as a valid protocol.

Practical Applications

The main practical advantage of the 3MT over the traditional CSMs protocol is in the

ability to determine CS accurately in a single exercise test. Furthermore, the establishment of CS


Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

from the speed-time profile of the all-out test does not require linear or nonlinear regression

analyses – simply averaging the speed of the last 30 s. Sensitivity to change in the CPM

parameter estimates has a significant practical application in monitoring adaptation to training.28

Moreover, power-duration based training intensity zones have been created using 3MT end

power to allow constant monitoring of training progress.36 Given that the approach from cycling

and running can be applied then swimming can employ these methods to provide enhanced

training monitoring and prescription. The current study has quantified the close relation between

3MT and traditional CSM derived estimates of CS. The appropriateness and efficacy of

employing training programs derived from these the 3MT estimates must be evaluated before

widespread adoption of this approach.

Conclusion

This study has demonstrated that the 3MT is a valid alternative protocol for swimming to

estimate aerobic capacity, CS. The demonstrated concurrent validity of the single session 3MT

might allow more widespread use of CSM to evaluate participants and responses to training. It is

recommended further investigations should focus on the relationship between D’ estimates

derived from 3MT and traditional CSMs to provide more accurate performance prediction and

applications for swimmers.


“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Acknowledgements

Thanks to all the subjects for volunteering to participate and coaches in assisting with data

collection in this study, in addition to, Ontario Ministry of Health and Long-Term Care for their

funding support. The funding agency had no role in the study design, collection, analysis and

interpretation of data, writing of the report, or in the decision to submit the article for

publication. The results of the current study do not constitute endorsement of the product by the

authors or the journal. None of the authors report any conflict of interest.
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Reference

1. Vanhatalo A, Doust JH and Burnley M. Determination of critical power using a 3-min all-
out cycling test. Medicine and science in sports and exercise. 2007; 39: 548-55.

2. Moritani T, Nagata A, deVries HA and Muro M. Critical power as a measure of physical


work capacity and anaerobic threshold. Ergonomics. 1981; 24: 339-50.

3. Brickley G, Dekerle J, Hammond AJ, Pringle J and Carter H. Assessment of maximal


aerobic power and critical power in a single 90-s isokinetic all-out cycling test.
International journal of sports medicine. 2007; 28: 414-9.

4. Dekerle J, Brickley G, Hammond AJ, Pringle JS and Carter H. Validity of the two-
parameter model in estimating the anaerobic work capacity. European journal of applied
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

physiology. 2006; 96: 257-64.

5. Burnley M, Doust JH and Vanhatalo A. A 3-min all-out test to determine peak oxygen
uptake and the maximal steady state. Medicine and science in sports and exercise.
2006; 38: 1995-2003.

6. Hughson RL, Orok CJ and Staudt LE. A high velocity treadmill running test to assess
endurance running potential. International journal of sports medicine. 1984; 5: 23-5.

7. Smith JC, Dangelmaier BS and Hill DW. Critical power is related to cycling time trial
performance. International journal of sports medicine. 1999; 20: 374-8.

8. Hill DW, Alain C and Kennedy MD. Modeling the relationship between velocity and time
to fatigue in rowing. Medicine and science in sports and exercise. 2003; 35: 2098-105.

9. Zagatto AM, Papoti M and Gobatto CA. Validity of critical frequency test for measuring
table tennis aerobic endurance through specific protocol. Journal of sports science &
medicine. 2008; 7: 461-6.

10. Toubekis AG and Tokmakidis SP. Metabolic responses at various intensities relative to
critical swimming velocity. Journal of strength and conditioning research / National
Strength & Conditioning Association. 2013; 27: 1731-41.

11. Wakayoshi K, Yoshida T, Udo M, et al. A simple method for determining critical speed as
swimming fatigue threshold in competitive swimming. International journal of sports
medicine. 1992; 13: 367-71.

12. Pettitt RW, Jamnick N and Clark IE. 3-min all-out exercise test for running. International
journal of sports medicine. 2012; 33: 426-31.

13. Cheng CF, Yang YS, Lin HM, Lee CL and Wang CY. Determination of critical power in
trained rowers using a three-minute all-out rowing test. European journal of applied
physiology. 2012; 112: 1251-60.

14. Kalva-Filho CA, Zagatto AM, Araujo MI, et al. Relationship between aerobic and
anaerobic parameters from 3-minute all-out tethered swimming and 400-m maximal front
crawl effort. Journal of strength and conditioning research / National Strength &
Conditioning Association. 2015; 29: 238-45.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

15. Wright B, Smith, D. A protocol for the determination of critical speed as an index of
swimming endurance performance. Medicine and Science in Aquatic Sports. 1994; 39:
55-9.

16. Takeda T, Ichikawa H, Takagi H and Tsubakimoto S. Do differences in initial speed


persist to the stroke phase in front-crawl swimming? Journal of sports sciences. 2009;
27: 1449-54.

17. di Prampero PE, Dekerle J, Capelli C and Zamparo P. The critical velocity in swimming.
European journal of applied physiology. 2008; 102: 165-71.

18. Zacca R, Wenzel BM, Piccin JS, Marcilio NR, Lopes AL and de Souza Castro FA.
Critical velocity, anaerobic distance capacity, maximal instantaneous velocity and
aerobic inertia in sprint and endurance young swimmers. European journal of applied
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

physiology. 2010; 110: 121-31.

19. di Prampero PE. The concept of critical velocity: a brief analysis. European journal of
applied physiology and occupational physiology. 1999; 80: 162-4.

20. Costa AM, Silva AJ, Louro H, et al. Can the curriculum be used to estimate critical
velocity in young competitive swimmers? Journal of sports science & medicine. 2009; 8:
17-23.

21. Pyne D, Trewin C and Hopkins W. Progression and variability of competitive


performance of Olympic swimmers. Journal of sports sciences. 2004; 22: 613-20.

22. Atkinson G, Edwards B, Reilly T and Waterhouse J. Exercise as a synchroniser of


human circadian rhythms: an update and discussion of the methodological problems.
European journal of applied physiology. 2007; 99: 331-41.

23. Reilly T and Marshall S. Circadian rhythms in power output on a swim bench. Journal of
Swimming Research. 1991: 11-3.

24. MacIntosh BR and Wright BM. Caffeine ingestion and performance of a 1,500-metre
swim. Canadian journal of applied physiology = Revue canadienne de physiologie
appliquee. 1995; 20: 168-77.

25. Vanata DF, Mazzino N, Bergosh R and Graham P. Caffeine improves sprint-distance
performance among division II collegiate swimmers. The Sport Journal. 2014.

26. Jenkins DG and Quigley BM. Blood lactate in trained cyclists during cycle ergometry at
critical power. European journal of applied physiology and occupational physiology.
1990; 61: 278-83.

27. Gaesser GA and Wilson LA. Effects of continuous and interval training on the
parameters of the power-endurance time relationship for high-intensity exercise.
International journal of sports medicine. 1988; 9: 417-21.

28. Vanhatalo A, Doust JH and Burnley M. Robustness of a 3 min all-out cycling test to
manipulations of power profile and cadence in humans. Experimental physiology. 2008;
93: 383-90.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

29. Gaesser GA, Carnevale TJ, Garfinkel A, Walter DO and Womack CJ. Estimation of
critical power with nonlinear and linear models. Medicine and science in sports and
exercise. 1995; 27: 1430-8.

30. Poole DC, Ward SA, Gardner GW and Whipp BJ. Metabolic and respiratory profile of the
upper limit for prolonged exercise in man. Ergonomics. 1988; 31: 1265-79.

31. Bergstrom HC, Housh TJ, Zuniga JM, et al. Differences among estimates of critical
power and anaerobic work capacity derived from five mathematical models and the
three-minute all-out test. Journal of strength and conditioning research / National
Strength & Conditioning Association. 2014; 28: 592-600.

32. Johnson TM, Sexton PJ, Placek AM, Murray SR and Pettitt RW. Reliability analysis of
the 3-min all-out exercise test for cycle ergometry. Medicine and science in sports and
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

exercise. 2011; 43: 2375-80.

33. de Lucas RD, Greco CC, Dekerle J, Carita RA, Guglielmo LG and Denadai BS. Test-
retest reliability of a 3-min isokinetic all-out test using two different cadences. Journal of
science and medicine in sport / Sports Medicine Australia. 2014; 17: 645-9.

34. Serresse O, Lortie G, Bouchard C and Boulay MR. Estimation of the contribution of the
various energy systems during maximal work of short duration. International journal of
sports medicine. 1988; 9: 456-60.

35. Jones MD, Booth J, Taylor JL and Barry BK. Aerobic training increases pain tolerance in
healthy individuals. Medicine and science in sports and exercise. 2014; 46: 1640-7.

36. Francis JT, Jr., Quinn TJ, Amann M and LaRoche DP. Defining intensity domains from
the end power of a 3-min all-out cycling test. Medicine and science in sports and
exercise. 2010; 42: 1769-75.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Linear Distance−Time Linear Speed−1/Time

C S = 1.11 m/s C S = 1.13 m/s

1.45
700

D = 28.59 m D = 24.5 m
r 2 = 0.9998 r 2 = 0.9925
Distance (m)

Speed (m/s)

1.35
500

1.25
300

1.15
100
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

100 200 300 400 500 600 700 0.002 0.006 0.010 0.014

Time (s) 1/Time (1/s)

Nonlinear Time−Speed 1.35 3−min all−out test


700

C S = 1.099 m/s C S = 1.1 m/s


D = 35.5 m D = 16.88 m
r 2 = 0.9957
500

1.25
Speed (m/s)
Time (s)

300

1.15
100

1.05

1.15 1.25 1.35 1.45 0 50 100 150

Speed (m/s) Time (s)

Figure 1. Curve fitting for each of the three CSM models and 3-min all-out test to estimate
critical speed and anaerobic distance capacity for one representative subject.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

10 20 30 40 50 A
D' (m)
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

3MT INV LIN NLIN

B
1.1
Speed (m/s)

0.9
0.7

3MT INV LIN NLIN

Models

Figure 2. Boxplot of D’ (A) and CS (B) for the CSMs and 3MT. (linear distance-time: LIN,
linear speed-time-1: INV, nonlinear time-speed: NLIN)
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

1.5 A B

50
y = 0.975x + 0.0333 y = 0.286x + 24.2
r = 0.9557 r = 0.3362

40
p = 0.00002 p = 0.342
LIN CS (m/s)

1.0

LIN D' (m)

30
20
0.5

10
0.0

0
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

0.0 0.5 1.0 1.5 0 10 20 30 40 50

3MT CS (m/s) 3MT D' (m)

C D
0.10

10
Differences in CS (m/s)

Differences in D' (m)

0
0.00

−10
−20
−0.10

−30

0.7 0.8 0.9 1.0 1.1 15 20 25 30

Average CS (m/s) Average D' (m)

Figure 3. Correlation and Bland-Altman analyses for differences in CS (A, C) and D’ (B, D)
estimates between 3MT and LIN. In panel A and B, the solid line is the best-fit linear regression
and the dashed line is the line of identity. In panel C and D, the solid horizontal line represents
the mean difference between the CS and D’ estimates, and the dashed lines represent the 95%
limits of agreement.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

1.5 A B

50
y = 0.963x + 0.0648 y = 0.338x + 17.2
r = 0.9514 r = 0.5153

40
p = 0.00002 p = 0.127
INV CS (m/s)

1.0

INV D' (m)

30
20
0.5

10
0.0

0
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

0.0 0.5 1.0 1.5 0 10 20 30 40 50

3MT CS (m/s) 3MT D' (m)

C 10 D
0.05
Differences in CS (m/s)

Differences in D' (m)

0
−0.05

−10
−20
−0.15

0.7 0.8 0.9 1.0 1.1 15 20 25 30

Average CS (m/s) Average D' (m)

Figure 4. Correlation and Bland-Altman analyses for differences in CS (A, C) and D’ (B, D)
estimates between 3MT and INV. In panel A and B, the solid line is the best-fit linear regression
and the dashed line is the line of identity. In panel C and D, the solid horizontal line represents
the mean difference between the CS and D’ estimates, and the dashed lines represent the 95%
limits of agreement.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

1.5 A B

50
y = 1.03x + −0.028 y = 0.421x + 27
r = 0.9588 r = 0.2323

40
p = 0.00001 p = 0.518
NLIN CS (m/s)

1.0

NLIN D' (m)

30
20
0.5

10
0.0

0
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

0.0 0.5 1.0 1.5 0 10 20 30 40 50

3MT CS (m/s) 3MT D' (m)

C D
0.10

20
Differences in CS (m/s)

Differences in D' (m)

0
0.00

−20
−40
−0.10

−60

0.7 0.8 0.9 1.0 1.1 15 20 25 30 35

Average CS (m/s) Average D' (m)

Figure 5. Correlation and Bland-Altman analyses for differences in CS (A, C) and D’ (B, D)
estimates between 3MT and NLIN. In panel A and B, the solid line is the best-fit linear
regression and the dashed line is the line of identity. In panel C and D, the solid horizontal line
represents the mean difference between the CS and D’ estimates, and the dashed lines represent
the 95% limits of agreement.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

1.5

*$

1.4

1.3
*$
*$
Speed (m/s)

1.2

*$
* *

1.1
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

0.9

0.8
15 30 45 60 75 90 105 120 135 150 165 180
* p < 0.05 compared to CS3MT Time Interval (s)
$ p<0.05 compared to the previous interval

Figure 6. The group mean speed averaged every 15 s during the 3MT.
“3-Min All-Out Test in Swimming” by Tsai MC, Thomas SC
International Journal of Sports Physiology and Performance
© 2016 Human Kinetics, Inc.

Table 1. Comparison of CS and D’ derived from 3MT with traditional CSMs. (linear distance-
time: LIN, linear speed-time-1: INV, nonlinear time-speed: NLIN)

CS Estimates D’ Estimates
Subject 3MT LIN INV NLIN 3MT LIN INV NLIN
(m s ) (m s ) (m.s-1) (m.s-1)
. -1 . -1
(m) (m) (m) (m)
1 0.94 0.92 0.94 0.91 9.68 29.66 23.77 36.05
2 0.71 0.70 0.72 0.68 20.85 36.38 27.34 48.71
3 0.94 0.96 0.96 0.95 27.32 35.68 34.38 37.16
4 0.91 0.91 0.93 0.90 13.22 26.60 20.66 34.81
5 1.02 0.97 0.98 0.98 8.33 19.46 18.34 14.93
6 0.74 0.82 0.82 0.80 32.33 26.26 23.02 33.16
7 0.94 0.98 0.98 1.00 17.31 21.46 22.00 8.92
Downloaded by Australian Catholic University on 09/24/16, Volume 0, Article Number 0

8 1.02 1.08 1.13 1.06 16.37 28.70 17.44 39.87


9 0.75 0.73 0.79 0.70 17.58 39.51 19.87 54.47
10 1.10 1.11 1.13 1.10 13.14 28.59 24.51 35.67
Mean 0.91 0.92 0.94* 0.91 17.51 29.23** 23.13 34.37*
SD 0.13 0.14 0.13 0.14 7.54 6.42 4.94 13.66
*
Significantly different from 3MT (p < 0.05)
**
Significantly different from 3MT (p < 0.001)

You might also like