Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/281710686

ANewModelforEstimatingPeakVO2BasedonPost-ExerciseMeasurementsinSwimming

ARTICLEinINTERNATIONALJOURNALOFSPORTSPHYSIOLOGYANDPERFORMANCE·SEPTEMBER2015

ImpactFactor:2.68·DOI:10.1123/ijspp.2015-0227

DOWNLOADS

27

VIEWS

65

5AUTHORS,INCLUDING:

DiegoChaverri

UniversityofBarcelona

19PUBLICATIONS17CITATIONS

SEEPROFILE

T.Schuller

DeutscheSporthochschuleKöln

8PUBLICATIONS1CITATION

SEEPROFILE

XavierIglesias

NationalInstituteofPhysicalEducation(INE…

77PUBLICATIONS111CITATIONS

SEEPROFILE

FerranA.Rodríguez

UniversityofBarcelona

170PUBLICATIONS1,017CITATIONS

SEEPROFILE

Availablefrom:FerranA.Rodríguez

Retrievedon:21September2015

For Peer Review

A Model for Estimating Peak Oxygen Uptake Based on Post-

Exercise Measurements in Swimming

Journal: International Journal of Sports Physiology and Performance

Manuscript ID: IJSPP.2015-0227.R2

Manuscript Type: Original Investigation

Keywords: exercise physiology, sport physiology, kinetics, aerobic

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

1

Submission Type: Original Investigation 1

Text-Only Word Count: 2510 2

Abstract Word Count: 222 3

Running Head: Peak VO2 estimation in swimming 4

Table: 1; Figures: 2 5

6

7

A New Model for Estimating Peak VO2 Based on Post-Exercise Measurements in 8

Swimming 9

10

Diego Chaverri1, Thorsten Schuller2, Xavier Iglesias1, Uwe Hoffmann2, Ferran A. 11

Rodríguez1 12

13

14 1 INEFC-Barcelona Sport Sciences Research Group, Institut Nacional d’Educació Física 15

de Catalunya, Universitat de Barcelona, Barcelona, Spain 16 2 Institut für Physiologie und Anatomie, Deutsche Sporthochschule Köln, Cologne, 17

Germany 18

19

20

21

22

23

24

� Prof. Ferran A. Rodríguez 25

Institut Nacional d’Educació Física de Catalunya (INEFC) 26

Universitat de Barcelona 27

Av. de l´Estadi 12-22, 08038 Barcelona, Spain. 28

Phone: + 34 93 425 54 45 29

E-mail: farodriguez@gencat.cat 30

Page 1 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

2

ABSTRACT 31

Purpose: Assessing cardiopulmonary function during swimming is a complex and 32

cumbersome procedure. Backward extrapolation is often used for predicting peak 33

oxygen uptake ( 2peakoV& ) during unimpeded swimming, but error can derive from a delay 34

at the onset of recovery. We assessed the validity of a mathematical model based 35

on heart rate (HR) and post-exercise kinetics for the estimation of 2peakoV& during 36

exercise. Methods: 34 elite swimmers performed a maximal front crawl 200-m swim. 37

was measured breath-by-breath and HR from beat-to-beat intervals. Data were time 38

aligned and 1-s interpolated. Exercise 2peakoV& was the average of the last 20 s of 39

exercise. Post-exercise was the first 20-s average during the immediate recovery. 40

Predicted values ( ) were computed using the equation:41

)HR(HR· )( o=)( o exercise-end22 ttVtVp && . Average values were calculated for different 42

time intervals and compared to measured exercise 2peakoV& . Results: Post-exercise 43

(0-20 s) underestimated 2peakoV& by 3.3% (95% CI = -9.8 under- to 3.2% overestimation; 44

mean difference = -116 ml·min-1; SEE = 4.2%; p = 0.001). The best 2peakoV& estimates 45

were offered by from 0-20 s (r2 = .96; mean diff. = 17 ml·min-1; SEE = 3.8%). 46

Conclusions: The high correlation (r2 = .86–-.96) and agreement between exercise and 47

predicted support the validity of the model, which provides accurate 2peakoV& 48

estimations after a single maximal swim whilst avoiding the error of backward 49

extrapolation and allowing the subject to swim completely unimpeded. 50

Key words: maximal oxygen uptake; oxygen kinetics; backward extrapolation; 51

modelling; heart rate 52

53

INTRODUCTION 54

The assessment of cardiopulmonary gas exchange and oxygen uptake ( ) in 55

swimming is a complex and cumbersome procedure and still faces limitations imposed 56

by the water environment and the equipment (see Sousa et al.1 for a review). 57

Specifically, in-water measurements require breathing through a snorkel connected with 58

a system of tubes and built-in valves that allows collecting the expired gases whilst 59

keeping dry the inspiratory and expiratory tubes, as well as the analysers. From a 60

technical standpoint, two main indirect calorimetric approaches have been used to 61

collect and analyse expiratory gases in swimming: (1) measurements during exercise 62

using snorkels with built-in valves connected to Douglas bags,2-4 open-circuit metabolic 63

charts,5,6 or breath-by-breath portable gas analysers;7 and (2) post-exercise 64

measurements with gas collection via facemask or mouthpiece connected to Douglas 65

bags8 or open-circuit metabolic charts.9 66

To enable continuous measurements in the field, portable gas analysers are now 67

preferred by many investigators because of their more advantageous sampling 68

capability, practicality, and acceptable level of accuracy.7,10 However, the inability for 69

swimmers to execute diving starts and underwater gliding after starts and turns, which 70

play a major role in the overall swimming performance, also impairs the ecological 71

validity of measurements. Even if these constraints do not prevent to investigatethe 72

investigation of many aspects of the physiological response during swimming, the 73

2oV&

2oV&

2oV&

2oV&

2oV& 2oVp &

2oV&

2peakOVp &

2oV&

2oV&

2oV&

Page 2 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

3

measurement of the respiratory function during exercise does restrict the full expression 74

of performance capacity in pool conditions, particularly during maximal swimming. For 75

instance, all-out 100-m front crawl and breastroke swims were ~5-6% slower when 76

using a snorkel compared to unimpeded swimming.11 An alternative procedure to 77

‘during-swimming’ measures is the backward extrapolation (BE) of the O2 recovery 78

curve, first described by di Prampero et al.12 and validated by Léger et al.13 Montpetit et 79

al. compared values obtained using the Douglas bag technique in a multistage 80

free-swimming test with those predicted using the BE method (i.e., recovery from the 81

same swimming test), as well as with those measured during an uphill treadmill running 82

test.8 Despite finding good method agreement, they concluded that, to ensure the 83

validity of the method, short duration exercises (<5 min) and supramaximal intensities 84

should be avoided as a delay in the onset of O2 recovery may appear. Another approach 85

was used by Costill et al.,14 who showed a good agreement between during 86

maximal and submaximal swimming and a single 20-s expired gas collection taken 87

immediately after a 4-7 min swim. However, breath-by-breath post-exercise 88

measurements confirmed the occurrence of a delay at the onset of the recovery 89

curve and identified a plateau in many —but not all— swimmers, suggesting this to be 90

the main source of error in these two estimation procedures.9,15 91

To overcome these limitations and to improve the estimation of from 92

post-exercise measurements, our group recently proposed a mathematical model based 93

on HR and off-transient kinetics.16 In short, based on the Fick’s principle, the 94

model calculates a predicted at a given time of recovery using the HR as a proxy 95

for changes in cardiac output, and the oxygen pulse as a proxy for the arteriovenous O2 96

difference.17 97

The aim of the present study was to assess the validity of this model by 98

comparing direct measurements during the final period of the swimming 99

exercise (reference method) with the one predicted by the model, as well as the one 100

indirectly estimated from a single 20-s measurement during recovery.14 Furthermore, 101

different recovery intervals were investigated in an attempt to enhance the accuracy of 102

estimation using the model. 103

104

METHODS 105

Subjects 106

Thirty-four elite swimmers, all members of national and Olympic teams, including 18 107

females (mean ± SD: age 20.8 ± 3.5 years, height 173.2 ± 5.8 cm, body mass 64.5 ± 5.6 108

kg) and 16 males (age 22.7 ± 3.6 years, height 186.8 ± 6.0 cm, body mass 80.8 ± 7.7 109

kg), voluntarily participated in this study. Informed consent was obtained from all 110

individual participants included in the study and their legal guardians when appropriate. 111

The study had received approval from the Ethics Committee for Clinical Sport Research 112

of Catalonia in accordance of the 1964 Helsinki declaration and its later amendments. 113

114

Design 115

All tests were conducted in a 50-m indoor pool (temperature: water 26–-27ºC, air 27–-116

28ºC). After a ~30 min swimming-based warm-up followed by 10 min of passive 117

recovery on pool side, participants completed an all-out 200-m front crawl swim to 118

determine exercise . After exercise, the swimmers remained still in an upright 119

peak2oV&

peak2oV&

2oV&

peak2oV&

2oV&

2oV&

peak2oV&

peak2oV&

peakV 2o&

Page 3 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

4

position immersed up to the sternum. In-water starts and touched open turns with no 120

underwater gliding were performed. 121

122

Methodology 123

was measured using a telemetric portable gas analyser (K4 b2, Cosmed, Italy) held 124

over the head of the swimmer by an assistant following the swimmers along the pool. 125

The equipment was connected to the swimmer by a low hydrodynamic resistance 126

respiratory snorkel and valve system, previously validated both in vivo18 and using a gas 127

exchange simulator.19 The gas analysers were calibrated before each test with gases of 128

known concentration (16% O2, 5% CO2) and the turbine volume transducer was 129

calibrated by using a 3-l syringe according to the manufacturer’s instructions. 130

Pulmonary gas exchanges were measured breath-by-breath 1 min before, during, and 3 131

min post-exercise. HR was continuously measured using waterproof beat-to-beat 132

monitors (CardioSwim, Freelap, Switzerland). 133

and HR data were time-aligned to the start of the measurements, 1-s 134

interpolated, and plotted against time. Two values were identified (figure 1): 1) 135

end-exercise was the average value over the last 20 s of exercise20 and was 136

taken as the reference value (criterion) for all comparisons; and 2) post-exercise 137

was the average value over the first 20 s of the recovery period ( ). The 20-s 138

duration of the end- and post-exercise measurements was chosen for the following 139

reasons: 1) to ensure that only last swimming lap data were used, as VO2 usually 140

decreases during the turns; 2) to minimize the influence of inter-breath fluctuations (i.e., 141

improvement of signal-to-noise ratio); 3) to prevent too high VO2max values frequently 142

obtained when using shorter time intervals;21 4) to maintain end- and post-exercise 143

temporal equality as previous results during 200-m maximal swimming showed an 144

on/off symmetry in the VO2 kinetic response;22 and 5) previous work showed that 20-s 145

average values produced the same VO2peak as the total amplitude of the 146

monoexponential equation fitting the VO2 on-kinetics during 200-m maximal swims.20 147

The rationale of the proposed new model relies on the Fick’s principle relating 148

cardiac output ( ) with and arterious-venous O2 difference ( ) 149

according to the equation: 150

&Vo2= &Q ⋅C(a− v )o

2 (1) 151

On the other hand, equals the cardiac stroke volume (SV) times the HR: 152

(2) 153

Under the assumption that SV does not significantly change over the first 154

seconds of recovery,23 changes in HR can be considered as a proxy for changes in , 155

and likewise, the /HR ratio can be used a proxy of the arterio-venous O2 difference: 156

constant ·o)v-a(≈HR

o2

2C

V& (3) 157

Based on these two assumptions, the mathematical model computes “predicted” 158

values ( ) based on synchronized post-exercise and HR measurements 159

(equation 3), and HR at the end of exercise. Thus, at a given time t during the recovery 160

period, can be calculated according to the equation: 161

2oV&

2oV&

peak2oV&

peak2oV&

2oV&

20)-(0o2V&

&Q 2oV& C(a− v )o2

&Q

&Q =SV ⋅HR

&Q

2oV&

2oV& 2oVp & 2oV&

2oVp &

Page 4 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

5

exercise-end2

2 HR)HR(

)( o=)( o

t

tVtVp

&& (4) 162

where, is the predicted(modelled) post-exercise 2oV& at time t; is the 163

post-exercise 1-s interpolated at time t; HR(t) is the post-exercise 1-s interpolated 164

HR value at time t; and HRend-exercise is the highest value of the last 10-s average HR at 165

the end of exercise (excluding single peaks higher than 5 bpm than the rest, 166

corresponding to ~1 SD from mean HR during the last 10 s of exercise). 167

In an attempt to enhance the accuracy of the estimation, was compared 168

with at different time intervals (t=0–20, 5–20, 10–20, 15–20, 5–15, and 10–15 s), 169

which were expressed as , , , , 170

, and . 171

172

Statistical analysis 173

Descriptive data are presented as mean, standard deviation (± SD), and mean difference. 174

Normality of distributions and homogeneity of variances were verified using the 175

Shapiro-Wilk’s and Levene’s tests, respectively. A one-way analysis of variance with 176

repeated measures (RM-ANOVA) and post-hoc Tukey's test if appropriate were used 177

for multiple comparisons between (criterion value) and each of the post-exercise 178

measured and predicted values. The Pearson's coefficient of determination (r2) was used 179

to assess correlation between variables. Mean difference plots24 were used to assess 180

agreement between measured and predicted values. The level of significance was set at 181

p < 0.05. Statistical analyses were conducted using SPSS 15.0 (SPSS Inc., Chicago, 182

USA). 183

184

RESULTS 185

Figure 1 shows an example of HR and over the last 30 s of the exercise and the 186

immediate recovery, as well as the predicted values for different post-exercise 187

intervals in one participant. In accordance with previous results,22,25 no evidence of a 188

slow component was observed in any swimmer as exercise duration constrained the 189

appearance of phase III of the kinetics.26 190

191

***Figure 1*** 192

193

Table 1 summarizes the comparisons between measured during exercise and 194

post-exercise measured and predicted values at different time intervals. End-exercise 195

(criterion value) was 3.3% higher than post-exercise . All predicted 196

were highly correlated with (r2 = 0.86–-0.96) and were not different from the 197

criterion value (p > 1.0-0.76–1.0). The best estimate (i.e. lowest bias) of exercise 198

was offered by (r2 = 0.96; mean diff. = 17 ml·min-1, SEE = 3.8%) 199

and (r2= = .94; mean diff. = 13 ml·min-1, SEE = = 4.7%). 200

201

***Table 1*** 202

203

)(o2 tVp & &Vo2(t)

2oV&

peak2oV&

2oVp &

)200(o2 −Vp & )205(o2 −Vp & )2010(o2 −Vp & )2015(o2 −Vp &

)155(o2 −Vp & )1510(o2 −Vp &

peak2oV&

2oV&

2oVp &

2oV&

peak2oV&

peak2oV& 20)-(0o2V&

2oVp &

peak2oV& )200(o2 −Vp &

)205(o2 −Vp &

Page 5 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

6

Figure 2 provides the Bland-Altman difference plots showing good agreement between 204

exercise and predicted recovery and data. 205

206

***Figure 2*** 207

208

209

DISCUSSION 210

Our main finding was that the proposed mathematical model based on HR kinetics and 211

post-exercise measurements in a maximal 200-m swim is a valid procedure of 212

estimating in competitive swimmers, with optimal predictive capacity if based 213

on measurements obtained during 20 s after the cessation of exercise. We also found 214

that assessing by using a single 20-s measurement during recovery as proposed 215

by Costill and coauthors14 is likely to underestimate true exercise by 3.3% on 216

average (95% CI =: 9.8 under- to 3.2% overestimation). 217

As previously explained, the model relies on the Fick’s principle (eq. 1 and 2), 218

and its basic assumption is that SV remains nearly constant during the first seconds of 219

recovery (eq. 3). The validity of this assumption needs further discussion. Following 220

light to moderate exercise, it has been shown that SV does not fall as rapidly as HR 221

does after exercise and remains above exercise levels23 for as long as 3 to 5 min,23,27,28 222

especially in the upright position.28 Sustained high during the recovery phase was 223

also demonstrated, explaining the on-/off-transient kinetics asymmetry (i.e. slower 224

off-transient time constant, also confirmed in 200-m maximal swimming22), appearing 225

to be a result of both SV and HR being maintained to ensure a sufficiently high O2 flow 226

to the muscle during recovery at a time when the muscle remains high.29 Therefore, 227

the decrease in (and consequently ) would occur mainly by decreased post-228

exercise HR. On the other hand, Sheldahl et al.30 suggested that the central 229

redistribution of blood volume with head-out water immersion cycling exercise at 40, 230

60 and 80% of 2maxoV& , leading to an increase in SV without a proportional decrease in 231

HR, evidences that is regulated at a higher level during upright exercise in water 232

compared with that on land. Although there is no available evidence that this response 233

pattern takes place during maximal exercise, in line with these reports, some of our 234

subjects showed a little rise in while HR remained constant immediately after the 235

cessation of exercise, which can be loosely interpreted as a rise in SV. This rise could 236

also be explained by the change in body position from horizontal to vertical; the lower 237

part of the body is now deep immersed and under a hydrostatic pressure gradient, which 238

would translocate blood from the lower limbs and abdomen into the thoracic region, 239

thus increasing venous return and SV compared to the horizontal position.31 We suggest 240

that this increase in SV is possibly the reason for the small overestimation of exercise 241

when shorter time intervals were used, such as in calculating and 242

(table 1). 243

A second assumption of the model is that remains nearly constant 244

during the first seconds of recovery. This assumption relies on the fact that a certain 245

venous volume with constant O2 saturation can be assumed to occur during the 246

immediate recovery while arterial saturation is constant. Because of the distance 247

peak2oV& )200(o2 −Vp & )205(o2 −Vp &

2oV&

peak2oV&

peak2oV&

peak2oV&

&Q

2oV&

2oV&

&Q 2oV&

&Q

2oV&

peak2oV& )155(o2 −Vp &

)1510(o2 −Vp &

C(a− v )o2

Page 6 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

7

between muscle and mouth, substantial changes in over the first seconds of 248

recovery are not to be expected, as shown by Drescher et al.17 As shown in figure 1,249

values do not decline over the first 20 s after the cessation of exercise, which is 250

likely the time for the onset of changes in perceivable in the exhaled air. On 251

the other hand, the high correlation and the low mean difference between during 252

exercise and strengthen the validity of the physiological assumptions of 253

the model. In the present findings, the off-kinetics isare virtually parallel to HR off- 254

kinetics during the first 20 s of recovery, suggesting that no substantial changes in 255

and SV occurred within this time period. Therefore, it seems justified to use 256

HR on- and off-kinetics as a proxy of dynamic response during the early recovery 257

within the limited scope of practical application of the model. 258

From another standpoint, the observation that and 259

provided the smallest estimation bias of during exercise is in agreement with 260

previous findings, showing a time-variable delay in the recovery curve after 261

maximal swimming.15 Our results show that this was likely to be the reason for (1) the 262

~20% overestimation of found by Lavoie et al. after an 400-m swim,32 and (2) 263

the similar results reported by Costill et al., who found a decline in during the first 264

20 s after the cessation of exercise causing a ~6% overestimation of after 4-7 min 265

of tethered swimming.14 Differences in methods of assessment and instrumentation 266

among these studies (e.g., Douglas bags vs. modern breath-by-breath oximeters) would 267

certainly explain at least some of these discrepancies. Concerning the variability of the 268

estimated parameters (figure 2, table 1), in which some predicted values deviate up to 269

~8% from measured values, we need to consider that the SEE for , the best 270

predictor variable, was 3.82%, very similar to the SEE for postexercise measured 271

(4.15%). This suggests that the main reason for these larger deviations is measurement 272

error, not inherent to the modelling procedure. Hyperventilation during the immediate 273

recovery appears to be the most straightforward explanation. 274

275

Practical applications 276

The proposed model allows to minimize the error in predicting from recovery 277

measurements after a maximal swim, with the practical advantage of avoiding the use of 278

respiratory equipment during swimming and, thus, allowing the swimmer to utilise their 279

normal breathing pattern and the full use of high-speed swimming technique and of the 280

specifically trained muscle mass in pool conditions. 281

From a technical standpoint, three conditions are required to ensure the validity 282

of the results: 1) obtaining quality beat-to-beat HR recordings, 2) obtaining the first 283

breath-by-breath values as fast as possible while avoiding missing breaths and 284

hyperventilation (e.g., as when swimmers are incorrectly advised to hold their breath 285

during the final strokes), and 3) monitoring HR and during a recovery period of at 286

least 20 s to avoid over- or underestimation. 287

Further validation of the model would imply comparing direct 2oV& 288

measurements with model-predicted values on swimming bouts of different duration 289

C(a− v )o2

2oVp &

C(a− v )o2

peak2oV&

)200(o2 −Vp &

2oV&

C(a− v )o2

2oV&

)200(o2 −Vp & )205(o2 −Vp &

peak2oV&

2oV&

peak2oV&

2oV&

2oV&

)200(o2 −Vp &

2oV&

peak2oV&

2oV&

2oV&

Page 7 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

8

and intensity (e.g., 100- to 400-m submaximal and maximal swims). In addition, more 290

basic studies investigating directly measured kinetics after maximal swimming 291

exercise would be required to confirm the physiological assumptions of the model. 292

293

CONCLUSION 294

We propose the new model, based on continuous beat-to-beat HR and post-exercise 295

breath-by-breath measurements during 20 s, as a valid and accurate procedure for 296

estimating in competitive swimmers. This calculation method avoids the bias of 297

estimations incurred by using the backward extrapolation method and 298

overcomes the constraints imposed by the use of respiratory equipment during 299

swimming. 300

301

ACKNOWLEDGEMENTS 302

This work was partially supported by grants awarded by the Ministry of Science and 303

Innovation of Spain (DEP2009-09181), Higher Sports Council of Spain (CSD 304

35/UPB/10, 005/UPB10/11, 112/UPB10/12, CAR-UGr 2009, CAR-UGr 2011), and 305

INEFC (Research Support Grants 2011, 2012). The contribution of Anna Barrero 306

(INEFC) in data acquisition is acknowledged. We are indebted to the coaches and staff 307

of the participating teams: Fred Vergnoux (C. N. Sabadell and RFEN); Jordi Murio, 308

Juan J. Castillo and Víctor Mancha (RFEN); David Lyles, Jenny Lyles and Xu Feng Jie 309

(Shanghai Province Swimming and Chinese Swimming Federation); Miha Potočnik, 310

Gorazd Podržavnik and Roni Pikec (Slovenian Swimming Federation); Rohan Taylor, 311

Jeremy Oliver and Danielle Stefano (Victorian Institute of Sport); and Patrick Pearson 312

(EIFFEL Swimmers PSV-Eindhoven). A very special note of appreciation goes to each 313

and all the swimmers who served as subjects and rendered their valuable time and 314

effort. 315

316

REFERENCES 317

1. Sousa A, Figueiredo P, Pendergast D, Kjendlie PL, Vilas-Boas JP, Fernandes 318

RJ. Critical evaluation of oxygen-uptake assessment in swimming. Int J Sports 319

Physiol Perform. Mar 2014;9(2):190-202. 320

2. Holmér I. Oxygen uptake during swimming in man. J Appl Physiol. Oct 321

1972;33(4):502-509. 322

3. Holmér I. Energy cost of arm stroke, leg kick, and the whole stroke in 323

competitive swimming styles. Eur J Appl Physiol Occup Physiol. 324

1974;33(2):105-118. 325

4. Kjendlie PL, Ingjer F, Madsen O, Stallman RK, Stray-Gundersen J. Differences 326

in the energy cost between children and adults during front crawl swimming. 327

Eur J Appl Physiol. Apr 2004;91(4):473-480. 328

5. Vilas-Boas JP, Santos P. Comparison of swimming economy in three 329

breaststroke techniques. In: Miyashita M, Mutoh Y, Richardson AB, eds. 330

Medicine and science in aquatic sports. Basel: Karger; 1994:48-54. 331

6. Fernandes RJ, Cardoso CS, Soares SM, Ascensão A, Colaço PJ, Vilas-Boas JP. 332

Time limit and VO2 slow component at intensities corresponding to VO2max in 333

swimmers. Int J Sports Med 2003;24(08):576-581. 334

7. Rodríguez FA, Keskinen KL, Keskinen OP, Malvela M. Oxygen uptake kinetics 335

during free swimming: a pilot study. In: Chatard J-C, ed. Biomechanics and 336

&Q

2oV&

peak2oV&

peak2oV&

Page 8 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

9

medicine in swimming IX. Saint-Étienne: Publications de l’Université de Saint-337

Étienne; 2003:379-384. 338

8. Montpetit RR, Léger LA, Lavoie JM, Cazorla G. VO2 peak during free 339

swimming using the backward extrapolation of the O2 recovery curve. Eur J 340

Appl Physiol Occup Physiol. 1981;47(4):385-391. 341

9. Rodríguez FA. Metabolic evaluation of swimmers and water polo players. 342

Kinesiology. Journal of Biology of Exercise. 1997 1997;2(1):19-29. 343

10. Fernandes RJ, Figueiredo P, Vilas-Boas JP. About the use and conclusions 344

extracted from a single tube snorkel used for respiratory data acquisition during 345

swimming. J Physiol Sci. Mar 2013;63(2):155-157. 346

11. Barbosa T, Silva AJ, Reis AM, Costa M, Garrido N, Policarpo F, Reis VM. 347

Kinematical changes in swimming front crawl and breaststroke with the 348

AquaTrainer snorkel. Eur J Appl Physiol. Aug 2010;109(6):1155-1162. 349

12. Di Prampero PE, Cortili G, Mognoni P, Saibene F. Energy cost of speec skating 350

and efficiency of work against air resistance. J Appl Physiol. Apr 351

1976;40(4):584-591. 352

13. Léger LA, Seliger V, Brassard L. Backward extrapolation of VO2max values 353

from the O2 recovery curve. Med Sci Sports Exerc. Spring 1980;12(1):24-27. 354

14. Costill DL, Kovaleski J, Porter D, Kirwan J, Fielding R, King D. Energy 355

expenditure during front crawl swimming: predicting success in middle-distance 356

events. Int J Sports Med. Oct 1985;6(5):266-270. 357

15. Rodríguez FA. Cardiorespiratory and metabolic field testing in swimming and 358

water polo: from physiological concepts to practical methods. In: Keskinen KL, 359

Komi PV, Hollander AP, eds. Biomechanics and medicine in swimming VIII. 360

University of Jyväskylä, Finland: Gummerus Printing; 1999:219-226. 361

16. Schuller T, Rodríguez FA, Iglesias X, Barrero A, Chaverri D, Hoffmann U. A 362

new model for estimating peak oxygen uptake based on post-exercise 363

measurements and heart rate kinetics in swimming. In: Balagué N, Torrents C, 364

Vilanova A, Cadefau J, Tarragó R, Tsolakidis E, eds. Proceedings of the 18th 365

Annual Congress of the European College of Sport Science. Barcelona: Open 366

Print; 2013:42. 367

17. Drescher U, Essfeld D, Hoffmann U. Modeling muscular V’O2-kinetics on the 368

basis of respiratory V’O2 and cardiac output measurements. In: Kurkusuz F, ed. 369

15th Annual Congress of the European College of Sport Science. Antalya: 370

Middle East Technical University; 2010:214. 371

18. Keskinen KL, Rodríguez FA, Keskinen OP. Respiratory snorkel and valve 372

system for breath-by-breath gas analysis in swimming. Scand J Med Sci Sports. 373

Oct 2003;13(5):322-329. 374

19. Rodríguez FA, Keskinen KL, Kusch M, Hoffmann U. Validity of a swimming 375

snorkel for metabolic testing. Int J Sports Med. Feb 2008;29(2):120-128. 376

20. Rodríguez FA, Iglesias X, Feriche B, Calderon-Soto C, Chaverri D, Wachsmuth 377

NB, Schmidt W, and Levine BD. Altitude training in elite swimmers for sea 378

level performance (Altitude Project). Med Sci Sports Exerc. 2015;47(9):1965-379

1978. 380

21. de Jesus K, Guidetti L, de Jesus K, Vilas-Boas JP, Baldari C, and Fernandes RJ. 381

Which are the best VO2 sampling intervals to characterize low to severe 382

swimming intensities? Int J Sports Med. Jun 2014;35(12):1030-1036. 383

22. Sousa A, Figueiredo P, Keskinen KL, Rodríguez FA, Machado L, Vilas-Boas 384

JP, Fernandes RJ. VO2 off transient kinetics in extreme intensity swimming. J 385

Sports Sci Med. 2011;10(3):546-552. 386

Page 9 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

10

23. Eriksen M, Waaler BA, Walloe L, Wesche J. Dynamics and dimensions of 387

cardiac output changes in humans at the onset and at the end of moderate 388

rhythmic exercise. J Physiol. Jul 1990;426:423-437. 389

24. Bland JM, Altman DG. Statistical methods for assessing agreement between two 390

methods of clinical measurement. The Lancet. 1986;327(8476):307-310. 391

25. Sousa AC, Figueiredo P, Oliveira NL, Oliveira J, Silva AJ, Keskinen KL, 392

Rodríguez FA, Machado LJ, Vilas-Boas JP, Fernandes RJ. VO2 kinetics in 200-393

m race-pace front crawl swimming. Int J Sports Med. Oct 2011;32(10):765-770. 394

26. Scheuermann BW, Barstow TJ. O2 uptake kinetics during exercise at peak O2 395

uptake. J Appl Physiol. Nov 2003;95(5):2014-2022. 396

27. Miyamoto Y, Higuchi J, Abe Y, Hiura T, Nakazono Y, Mikami T. Dynamics of 397

cardiac output and systolic time intervals in supine and upright exercise. J Appl 398

Physiol Respir Environ Exerc Physiol. Dec 1983;55(6):1674-1681. 399

28. Takahashi T, Hayano J, Okada A, Saitoh T, Kamiya A. Effects of the muscle 400

pump and body posture on cardiovascular responses during recovery from cycle 401

exercise. Eur J Appl Physiol. Aug 2005;94(5-6):576-583. 402

29. Yoshida T, Whipp BJ. Dynamic asymmetries of cardiac output transients in 403

response to muscular exercise in man. J Physiol. Oct 15 1994;480 ( Pt 2):355-404

359. 405

30. Sheldahl LM, Tristani FE, Clifford PS, Hughes CV, Sobocinski KA, Morris RD. 406

Effect of head-out water immersion on cardiorespiratory response to dynamic 407

exercise. J Amer Coll Cardiol. 1987;10(6):1254-1258. 408

31. Arborelius M, Jr., Ballidin UI, Lilja B, Lundgren CE. Hemodynamic changes in 409

man during immersion with the head above water. Aerosp Med. Jun 410

1972;43(6):592-598. 411

32. Lavoie J-M, Léger LA, Montpetit RR, Chabot S. Backward extrapolation of 412

VO2 from the O2 recovery curve after a voluntary maximal 400m swim. In: 413

Hollander AP, Huijing PA, De Groot G, eds. Biomechanics and medicine in 414

swimming: Human Kinetics, Champaign, Illinois.; 1983:222-227. 415

Page 10 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

11

FIGURE LEGENDS 416

417

Figure 1. Heart rate (light grey diamonds) and (dark grey squares) kinetics over the 418

last 30 s of exercise and immediate recovery during a 200-m maximal swim in one 419

swimmer. Vertical lines indicate time intervals during exercise (t < 0) and recovery (t > 420

0). In black circles, modelled (predicted) values ( ) during recovery. Short dashed 421

lines indicate different time intervals used for comparisons. 422

423

Figure 2. Bland-Altman difference plots between peak at the end of exercise 424

(criterion) and the two best estimates calculated by the model, (panel A), 425

and (panel B), respectively. The equality (solid), mean difference (long-426

dashed), and ± 95% limits of agreement (short-dashed) lines are depicted. 427

428

429

2oV&

2oVp &

2oV&

)200(o2 −Vp &

)205(o2 −Vp &

Page 11 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

260x190mm (150 x 150 DPI)

Page 12 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

529x396mm (72 x 72 DPI)

Page 13 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance

For Peer Review

Table 1. Oxygen uptake measurements during exercise (criterion value) and recovery (post-exercise), and values predicted by the model at

different time intervals.

2oV&

Time interval 95% CI Mean diff. r2 SEE Significance

(s) (ml·min-1

) (ml·min-1

) (ml·min-1

) (ml·min-1

) (%) (p-value)

Exercise (criterion) -20–0 3547 ± 692 3305 3788 0 – – – –

Post-exercise 0–20 3431 ± 685 3192 3670 -116 .959 142 4.15 .001*

Predicted 0–20 3564 ± 698 3320 3807 17 .963 136 3.82 1.0

5–20 3559 ± 705 3313 3805 13 .943 168 4.72 1.0

10–20 3520 ± 725 3267 3773 -27 .900 222 6.30 1.0

15–20 3438 ± 722 3186 3690 -109 .856 267 7.76 .76

5–15 3623 ± 707 3376 3869 76 .963 135 3.74 .07

10–15 3604 ± 731 3349 3859 57 .923 195 5.42 1.0

Values are mean ± SD. 95% CI, 95% confidence interval; Mean diff., mean difference with criterion value; r2, Pearson’s coefficient of

determination; SEE, standard error of estimate; Significance, compared with criterion value; *Significantly different from criterion value (p

< 0.05).

2oV&

Page 14 of 14

Human Kinetics, 1607 N Market St, Champaign, IL 61825

International Journal of Sports Physiology and Performance