ORIGINAL ARTICLE

Circulatory ‘‘Efficacy’’ during progressive aerobic exercisein children: insights from the Q:VO2 relationship

Thomas Rowland

Accepted: 11 April 2007 / Published online: 3 May 2007

� Springer-Verlag 2007

Abstract The relationship between circulatory flow (Q)

and oxygen uptake (VO2) may provide insights into per-

formance of peripheral mechanisms which govern blood

flow during exercise (circulatory efficacy). This study

evaluated the response of Q relative to VO2 during pro-

gressive upright cycle exercise in a group of 39 preado-

lescent boys (mean age 12.2 ± SD 0.5 years). The Q–VO2

relationship was curvilinear, best described by the cubic

equation Q = 3.60(VO2)3 + 5.24(VO2)2 + 2.40(VO2) –

0.94. Circulatory efficacy, defined as the %DQ/

%DVO2 · 100, fell from 70.4% between the first two

workloads to 49.7% at peak exercise. This decline in cir-

culatory efficacy is consistent with other published data

suggesting a decline in skeletal muscle pump function at

high intensity workloads. The pattern of change in rela-

tionship of Q and VO2 during progressive exercise in these

children is similar to that observed in studies of adults. This

implies that performance of peripheral determinants of

circulatory responses to exercise is not affected by bio-

logical maturation.

Keywords Cardiac output � Cardiovascular fitness �Oxygen uptake

Introduction

Circulatory responses to endurance exercise are mediated

primarily via peripheral mechanisms, particularly arteriolar

dilatation and skeletal muscle pump function (Tschakovsky

et al. 1996; Rowland 2005a). Traditional tenets hold that

during progressive exercise this circulatory response (Q) is

closely coupled with metabolic demand, or oxygen uptake

(VO2) (Astrand et al. 1964; Faulkner et al. 1977). It is

clear, however, that while Q generally rises proportion-

ately to increases in VO2, circulatory supply falls far short

of satisfying the oxygen requirements of exercising muscle

cells. In fact, for any given increase in work rate, oxygen

delivery by the circulation accounts for only approximately

50% of the rise in metabolic demand (Rowland 2001). To

compensate for this perfusion-demand gap, skeletal muscle

extracts an increasing amount of oxygen from each unit

volume delivered to the capillary-cell interface to satisfy

mitochondrial aerobic demands. Consequently, arterial

venous oxygen difference steadily rises as work intensity

increases.

This observation may bear importance in characterizing

the peripheral determinants of circulatory responses to

exercise, particularly the function of the skeletal muscle

pump (SMP). Gotshall et al. (1996) viewed the relative

contribution of Q to a given level of VO2 as a marker of the

efficacy of these responses. That is, according to this

concept, the ratio of Q/VO2 serves as a marker of the

performance of peripheral determinants of circulatory flow,

while arterial venous oxygen difference becomes a nega-

tive index of the adequacy of the circulatory response to

exercise.

It is important to recognize, as well, that evidence exists

that the Q–VO2 relationship is not entirely linear through

the course of a progressive exercise test (Grimby et al.

1966; Yamaguchi et al. 1986; Vella and Robergs 2005).

For example, Vella and Robergs (2005) recently described

a curvilinear relationship between Q and VO2, which

indicated a decrease in Q–VO2 slope of more than 1 l min–1

Q/1 l min–1 VO2 over a VO2 span of 4 l min–1. Others have

T. Rowland (&)

Department of Pediatrics, Baystate Medical Center,

Springfield, MA 01199, USA

e-mail: Thomas.rowland@bhs.org

123

Eur J Appl Physiol (2007) 101:61–66

DOI 10.1007/s00421-007-0472-1

reported a plateau in Q at high work intensities (Yamag-

uchi et al. 1986). These observations imply a decline in

circulatory efficacy—and, by inference, reduced effec-

tiveness of arteriolar dilatation and/or SMP function—as

maximal exercise level is approached.

This conclusion is consistent with recent work by Lut-

jemeier et al. (2005) indicating a decrease in skeletal

muscle pump function at high work intensities due to

increasing intramuscular vascular compression and inflow

occlusion. Others have indicated evidence of limited SMP

activity at maximal work loads (Takahashi and Miyamoto

1998; Rowland and Lisowski 2003). This information

suggests that peripheral factors other than SMP function

(i.e. arteriolar dilatation) may act to limit circulatory flow

in a progressive exercise test.

This study examined the relationship between Q and

VO2 during a standard progressive upright cycle test in a

group of nontrained preadolescent boys using Doppler

echocardiography. Clarifying the nature of this association

in the pediatric age group bears particular importance, as

children characteristically demonstrate a lower Q/VO2

during exercise than adults (Bar-Or 1983). Some have

interpreted this finding as indicative of a ‘‘hypokinetic’’

cardiac response to exercise in young subjects (Faulk

2000). Others have concluded the lower Q/VO2 in children

to be spurious and biologically irrelevant, since (1) children

and adults do not exercise at the same absolute VO2 and (2)

the ratio is explained simply by the lower stroke volume

expected in children (Rowland 2005b). An examination of

the course of the Q/VO2 relationship during progressive

exercise would help resolve these conflicting interpreta-

tions.

Methods

Thirty-nine boys (mean age 12.2 ± 0.5 years) from the

same sixth grade class agreed to perform maximal up-

right cycle exercise with measurements of gas exchange

and cardiovascular variables. Subjects were invited from

quartiles of performance on a standard school one-mile

run/walk and thus represented a full spectrum of car-

diovascular fitness. Data from these boys has previously

been published in reports of the physiologic and

anthropometric determinants of field and laboratory aer-

obic fitness (Rowland et al. 1999a, b) and also in a study

of allometric scaling of one-mile run performance (Nevill

et al. 2004).

The subjects were in good health and none had signifi-

cant obesity. Two-thirds had recently participated on a

community sports team, but none were involved in regular

athletic training. By questionnaire, 14 had onset of voice

change and/or pubic hair indicative of early puberty.

Subjects were asked to refrain from vigorous physical

activity within the 24 h before the exercise test. In an at-

tempt to create some homogeneity of fluid balance, each

boy drank 240 ml of a sports drink one hour before

appearing at the exercise laboratory. Height and weight

were measured with a stadiometer and calibrated balance

beam scale, respectively. Right-sided scapular and triceps

skinfold thicknesses were averaged from triplicate mea-

surements to the nearest 0.1 mm, and body fat was esti-

mated by the equations of Slaughter et al. (1988).

Subjects performed a continuous multi-stage cycle test

to exhaustion in the upright position in an air-conditioned

laboratory (19–21�C). Initial and incremental loads were

25 W, with 3 min stages and a steady pedaling cadence of

60 rpm. Subjects were verbal encouraged to achieve an

exhaustive effort, and the test was terminated when the

pedaling cadence could no longer be maintained. Peak

work capacity (PWC) was defined as the greatest work load

achieved, prorated for incomplete stages.

Heart rate was recorded by an electrocardiogram. Gas

exchange variables were measured by standard open circuit

techniques using a Q-Plex Cardiopulmonary Exercise

System (Quinton Instrument Company, Seattle, WA).

Subjects breathed through a Rudolph valve (94 ml dead

space) into a 6 l mixing chamber from which gas samples

were drawn for analysis (oxygen and carbon dioxide con-

tent by zirconia oxide and infrared analyzers, respectively).

Minute ventilation was measured by a pneumotachometer

in the expiratory line. The system was calibrated before

and after each test with standard gases of known oxygen

and carbon dioxide concentration.

Gas exchange values, including VO2, were averaged and

recorded at 15 s intervals. Submaximal VO2 levels were

determined as the average of values over the third minute

of each work stage, while peak VO2 was defined as the

mean of the two highest values in the final minute of

exercise. Peak VO2 was assumed to reflect VO2max with

peak heart >185 bpm and peak respiratory exchange ratio

(VCO2/VO2) >1.00, accompanied by subjective evidence

of fatigue (hypernpea, flushing or pallor, sweating).

Stroke volume was estimated at rest and during sub-

maximal and maximal exercise by standard Doppler

echocardiographic techniques (Rowland and Obert 2002).

Velocity in the ascending aorta was measured by a

1.9 MHz (Pedof) transducer directed inferiorly from the

suprasternal notch. Integration of velocity versus time

(VTI) was averaged for the 5–10 beats with highest con-

sistent values and stroke volume was estimated by multi-

plying this value times the aortic cross-sectional area (sino-

tubular junction) measured by two-dimensional echocar-

diography in the seated position at rest. Stroke volume was

determined at rest, in the final minute of each workload,

and in the final 30–60 s of exercise. Cardiac output was

62 Eur J Appl Physiol (2007) 101:61–66

123

calculated as the product of stroke volume and simulta-

neously measured heart rate. Circulatory efficacy was

examined quantitatively by two definitions: (1) the ratio of

scope (multiple of resting value) of Q versus VO2, and (2)

the percent change in Q relative to percent change in VO2

between successive workloads, expressed as a percent.

(i.e., the extent that change in VO2 could be accounted for

by change in Q). Arterial venous oxygen difference was

calculated as VO2/Q.

Reports of the validity and reliability of the Doppler

echocardiographic technique for estimating cardiac output

during maximal exercise have been previously published

from this laboratory (Rowland and Popowski 1997; Row-

land et al. 1998). This method is particularly useful for this

investigation, as Q and VO2 can be obtained simulta-

neously, measurement is nonobtrusive, and steady state is

not necessary.

The configuration of the relationship of Q to VO2 with

increasing workloads was examined by curve fitting using

commercial statistical software (SPSS 11.0 for Windows,

SPSS, Inc., Chicago, IL). Comparisons were made using

linear, cubic, quadratic, log, and exponential models,

defining the best fit as the model with the highest coeffi-

cient of determination (r2).

Results

Average weight and height of the subjects was

45.6 ± 10.1 kg and 153 ± 9 cm, respectively, with a mean

19.4 ± 6.5% estimated body fat. Two subjects did not sat-

isfy criteria for VO2max, but by staff observations both had

achieved a true exhaustive effort, and their data are included

in this analysis. Average VO2max was 47.0 ± 5.8 ml kg–1

min–1 with a mean PWC of 135 ± 21 W. Mean values of

maximal stroke index (related to body surface area) and

cardiac index were 61 ± 11 ml per m2 and 11.98 ± 2.33 l

min–1 per m2, respectively. Calculated arterial venous

oxygen difference rose progressively from a resting value of

5.8 ± 1.6 ml 100 ml–1 to 13.0 ± 2.5 ml 100 ml–1 at peak

exercise.

Figure 1 indicates a curvilinear relationship of Q and

VO2 during progressive exercise. The greatest r2 (1.00)

was defined by a cubic model, with the equation

Q = 3.60(VO2)3 + 5.24(VO2)2 + 2.40(VO2) – 0.94. The

linear equation (r2 = .989) was Q = 3.86VO2 + 6.24.

By both definitions, circulatory efficacy fell as exercise

intensity increased. Scope values for submaximal work-

loads 1–4 (25–100 W) and maximum (135 ± 12 W) were

1.49, 1.84, 2.21, 2.60, and 3.04 for Q, respectively, and

2.29, 3.06, 4.03, 5.10, and 6.87 for VO2. As demonstrated

in Fig. 2, this indicated a decline in circulatory efficacy

that reached <50% at high work loads.

Table 1 indicates submaximal and maximal data for Q

and VO2. Efficacy defined as %D Q relative to %D VO2

between successive work loads fell from 70.4% at light

work to 49.7% at high work intensity.

Discussion

Abundant evidence has accumulated supporting a critical

role for peripheral mechanisms (arteriolar dilatation, skel-

etal muscle pump function) in facilitating and controlling

circulatory responses to exercise (Rowland 2001, 2005a).

Delineating the functional characteristics of these factors,

2.22.01.81.61.41.21.00.80.66

7

8

9

10

11

12

13

14

15

16

17

QL min-1

VO2 L min-1

Fig. 1 Cardiac output (Q) ± standard deviation plotted relative to

oxygen uptake (VO2) during progressive exercise to exhaustion. Data

are presented at 25, 50, 75, 100, and 135 ± 21 W (max)

6.05.04.03.02.01.00

1.0

2.0

3.0

4.0

5.0

6.0

Q

50%

100%

VO2

Fig. 2 Scope of VO2 and Q (expressed as multiples of resting values)

with progressive exercise. The solid line of identity indicates 100%

circulatory efficacy (i.e., identical slope), while the dashed linedenotes 50% efficacy

Eur J Appl Physiol (2007) 101:61–66 63

123

however, has proven challenging. A number of experi-

mental approaches have been utilized, including compari-

son of exercise Q to flow with pathologic arterial venous

fistulae (Binak et al. 1960) and use of pharmacologic

vasodilators (Panchev et al. 2005) (to simulate local arte-

riolar dilatation during exercise conditions) and muscle

compression (Tschakovsky et al. 1996), electrical muscle

stimulation (Raymond et al. 1999), effect of changes in

cycling cadence and force (Sheriff 2002), and immediate

post-exercise flow measurements (Lutjemeier et al. 2005)

to assess skeletal muscle pump function.

As examined in the present study, the relationship be-

tween Q and VO2 may be interpreted as an indicator of

circulatory efficacy during exercise, and, by extension, a

marker of arteriolar dilatation and/or skeletal muscle pump

function. The findings in this group of 12-year old boys

using Doppler ultrasound indicated a non-linear relation-

ship between Q and VO2 during progressive exercise,

consistent with diminishing circulatory efficacy as muscle

contractile force and metabolic demand intensified. As a

consequence, efficacy (defined as the magnitude of in-

creased oxygen demand satisfied by a rise in Q) fell pro-

gressively during exercise.

This finding is consistent with recent reports in adult

subjects as well as earlier research information (see Clau-

sen 1976, for review). Vella and Robergs (2005) described

a curvilinear relationship between Q and VO2 in trained

adult male athletes throughout the duration of a progressive

upright cycling test while using the carbon dioxide reb-

reathing method for measuring Q. Using the dye dilution

technique in untrained adult men, Yamaguchi et al. found a

linear association between Q and VO2 in the submaximal

exercise intensity range (1986). However, above ~70%

VO2max the majority of subjects demonstrated a tapering

of Q and decreasing Q/VO2. On the other hand, others have

indicated no change in Q/VO2 during progressive exercise

in adults when Q was estimated by impedance cardiogra-

phy (Miyamoto et al. 1992) or carbon dioxide rebreathing

(Faulkner et al. 1977). McCole et al. (2001), utilizing the

acetylene rebreathing method, noted that demonstration of

a dissociation beween Q and VO2 at maximal exercise was

dependent on the duration of the testing protocol.

While the peripheral factors explaining this decline in

efficacy remain to be clarified, recent observations by

Lutjemeier et al. (2005) of a decline in SMP function at

high workloads may be particularly pertinent. Evidence

indicates that net performance of the SMP depends on the

balance of (1) enhanced flow from venous compression

(assisted by venous valves) and the negative suction effect

created by muscle contraction, and (2) inhibition of inflow

caused by the high pressures accompanying vascular

compression (Folkow et al. 1970; Sheriff et al. 1993). By

comparing muscle blood flow during and immediate after

exercise, Lutjemeier and her colleagues were able to con-

clude that the negative effect of compression became rel-

atively greater than flow enhancement as work intensity

increased. That is, at high work rates, mechanisms of en-

hanced flow by the SMP were not capable of compensating

for impairment of inflow due to the increasing force of

muscle contraction.

That the SMP plays a limited role in Q at peak exercise

is supported by other investigations. In a study involving

both men and boys, Rowland and Lisowski (2003) found

that cardiac output fell only 10–15% in the first 15–20 s

after the SMP was abruptly stopped at maximal or sub-

maximal exercise. Similarly, Takahashi and Miyamoto

(1998) reported a decrease in average cardiac output from

12.7 to 11.1 l min–1 in the first 20 s after submaximal

upright cycling in adult men. These observations would

suggest that the decline in Q/VO2 at higher work intensities

documented in this and other studies may reflect a decline

in circulatory efficacy resulting from decreased contribu-

tion of SMP.

It should be recognized that central (i.e. cardiac) as well

as peripheral factors can influence values of Q/VO2. Pa-

tients with congestive heart failure, for instance, exhibit

lower Q/VO2 during exercise (Clausen 1976). In this study

of healthy children, however, it would seem unlikely that

cardiac factors would have a negative influence on Q/VO2.

This study also provided an opportunity to assess the

possible effect of biological maturation on circulatory

efficacy during exercise. Bar-Or (1983) first brought

attention to the fact that any given absolute VO2 during

exercise is met with a lower Q in children compared to

adults. Combining data from several studies, he demon-

strated that such values of Q in children clustered at the

bottom of the normal range for adults as VO2 rose.

This ‘‘hypokinetic’’ cardiovascular response to exercise

by children has been considered to bear implications for

aerobic fitness in youth, particularly in respect to matura-

tional differences in performance in hot ambient tempera-

tures (Faulk 2000). Others have considered this observation

biologically spurious, an expression of the child’s obvi-

Table 1 Mean (standard deviation) values of cardiac output (Q) and

oxygen uptake (VO2) during progressive exercise

Q (l min–1) VO2 (l min–1) Efficacy (%)

25 W 8.11 (1.41) 0.71 (0.08)

50 W 10.04 (1.73) 0.95 (0.09) 70.4

75 W 12.08 (1.95) 1.25 (0.11) 64.2

100 W 14.15 (2.51) 1.58 (0.12) 64.8

Max (135 ± 21 W) 16.60 (3.48) 2.13 (0.41) 49.7

Efficacy = %D Q relative to %D VO2 between individual stages,

expressed as a percent

64 Eur J Appl Physiol (2007) 101:61–66

123

ously smaller stroke volume, with no physiologic or per-

formance importance (Rowland 2005b). This latter argu-

ment has been supported by data indicating that when the

response of cardiovascular variables to exercise are related

appropriately to body size, no differences are observed

between children and adults (Rowland et al. 1997; Nottin

et al. 2002).

The present study supports the latter viewpoint. As

indicated in Fig. 3, the observed curvilinear pattern of

decline of Q relative to VO2 with progressive exercise was

similar to that previously reported in adult subjects. (Dif-

ferences in absolute values in these different studies may

be explained by variations in measurement technique and

body position.) By inference, then, the magnitude of de-

cline in circulatory efficacy with increasing work intensity

is no different in children than adults.

In summary, the findings in this exercise study of 12-

year old boys using Doppler ultrasound indicate a non-

linear relationship between Q and VO2, consistent with the

concept of diminishing circulatory efficacy as work inten-

sifies. This reduced efficacy may reflect a reduction of

skeletal muscle pump performance at high work loads. In

addition, this investigation failed to reveal evidence of

maturational differences in the pattern of circulatory re-

sponses to progressive endurance exercise.

References

Astrand PO, Cuddy TR, Saltin B, Stenberg J (1964) Cardiac output

during submaximal and maximal work. J Appl Physiol 19:268–

274

Bar-Or O (1983) Pediatric sports medicine for the practitioner.

Springer, New York, p 20

Binak K, Regan TJ, Christensen RC (1960) Arteriovenous fistula:

hemodynamic effects of occlusion and exercise. Am Heart J

60:495–502

Clausen JP (1976) Circulatory adjustments to dynamic exercise

and effect of physical training in normal subjects and in

patients with coronary artery disease. Prog Cardiovasc Dis

17:459–495

Faulk B (2000) Temperature regulation. In: Armstrong N, van

Mechelen (eds) Paediatric exercise science and medicine.

Oxford University Press, Oxford, pp 221–239

Faulkner JA, Heigenhauser GJ, Schork MA (1977) The cardiac

output-oxygen uptake relationship of men during graded bicycle

ergometry. Med Sci Sports 9:148–154

Folkow B, Gaskell P, Waaler BA (1970) Blood flow through limb

muscles during heavy rhythmic exercise. Acta Physiol Scand

80:61–72

Gotshall RW, Bauer TA, Fahrner SL (1996) Cycling cadence alters

exercise hemodynamics. Int J Sports Med 17:17–21

Grimby G, Nilsson NJ, Saltin B (1966) Cardiac output during

submaximal and maximal exercise in active middle-aged

athletes. J Appl Physiol 21:1150–1156

Lutjemeier BJ, Miura A, Scheuermann BW, Koga S, Townsend DK,

Barstow TJ (2005) Muscle contraction-blood flow interactions

during upright knee extension exercise in humans. J Appl

Physiol 98:1575–1583

McCole SD, Davis AM, Fueger PT (2001) Is there a dissociation of

maximum oxygen consumption and maximal cardiac output?

Med Sci Sports Exerc 33:1265–1269

MiyamotoY, Niizeki K, Abe H (1992) Behavior of cardiac output

during progressive exercise tests: a preliminary report. Am

Physiol Anthrop 11:225–230

Nevill A, Rowland T, Goff D, Martel L, Ferrone L (2004) Scaling or

normalizing maximum oxygen uptake to predict 1-mile run time

in boys. Eur J Appl Physiol 92:285–288

Nottin S, Vinet A, Stecken F, Agnes V, N’Guyen LD, Ounissi F,

Lecoq AM, Obert P (2002) Central and peripheral cardiovascular

adaptations during maximal cycle exercise in boys and men.

Med Sci Sports Exerc 33:456–463

Panchev VS, Suvandjieva AV, Pancheva MV (2005) The muscle

pump is not an important determinant of muscle blood flow

during exercise. J Appl Physiol 99:778–783

Raymond J, Davis GM, Bryant G, Climdern M, Sutton JR (1999)

Cardiovascular responses to an orthostatic challenge and elec-

trical-stimulation-induced leg muscle contractions in individuals

with paraplegia. Eur J Appl Physiol 80:205–121

Rowland T (2001) The circulatory response to exercise: role of the

peripheral pump. Int J Sports Med 22:558–565

Rowland T (2005a) Circulatory responses to exercise. Are we

misreading Fick? Chest 127:1023–1030

Rowland T (2005b) Children’s exercise physiology. Human Kinetics.

Champaign, IL, p 114

Rowland T, Lisowski R (2003) Determinants of diastolic cardiac

filling during exercise. J Sports Med Phys Fitness 43:380–385

Rowland T, Obert P (2002) Doppler echocardiography for the

estimation of cardiac output with exercise. Sports Med 32:973–

986

Rowland T, Popowski B (1997) Comparison of bioimpedance and

Doppler cardiac output during exercise in children [abstract].

Pediatr Exerc Sci 9:188

Rowland T, Popowski B, Ferrone L (1997) Cardiac responses to

maximal upright cycle exercise in healthy boys and men. Med

Sci Sports Exerc 29:1146–1151

Rowland T, Melanson E, Popowski B (1998) Test-retest reproduc-

ibility of maximal cardiac output by Doppler echocardiography.

Am J Cardiol 81:1228–30

Rowland T, Kline G, Goff D, Martel L, Ferrone L (1999a) One-mile

run performance and cardiovascular fitness in children. Arch

Pediatr Adolesc med 153:845–849

2.22.01.81.61.41.21.00.80.65

7

9

11

13

15

17

19

21

23

QL min-1

VO2 L min-1

2.4 2.6

Grimby et al.

Yamaguchi et al.

Present Study

Vella & Robergs

Fig. 3 Pattern of change in Q relative to VO2 during progressive

exercise reported in previous studies of adults compared to that in the

present study of preadolescent boys

Eur J Appl Physiol (2007) 101:61–66 65

123

Rowland T, Kline G, Goff D, Martel L, Ferrone L (1999b)

Physiological determinants of maximal aerobic power in healthy

12-year old boys. Pediatr Exerc Sci 11:317–326

Sheriff DD (2002) Muscle pump function during locomotion:

mechanical coupling of stride frequency and muscle blood flow.

Am J Physiol H2185–H2191

Sheriff DD, Powell LB, Scher AM (1993) Is rapid rise in vascular

conductance at onset of dynamic exercise due to muscle pump?

Am J Physiol 265:H1227–H1234

Slaughter MH, Lohman TG, Boileau RA, Stillman RJ, Van Loan M,

Horswill CA, Wilmore JH (1988) Skinfold equations for

estimation of body fatness in children and youth. Hum Biol

60:709–723

Takahashi T, Miyamoto Y (1998) Influence of light physical activity

on cardiac responses during recovery from exercise in humans.

Eur J Appl Physiol 77:305–311

Tschakovsky ME, Shoemaker JK, Hughson RL (1996) Vasodilation

and muscle pump contribution to immediate exercise hyperemia.

Am J Physiol Heart Circ Physiol 271:H1697–H1701

Vella CA, Robergs RA (2005) Non-linear relationship between

central cardiovascular variables and VO2 during incremental

cycle exercise in endurance-trained individuals. J Sports Med

Phys Fitness 45:452–459

Yamaguchi I, Komatsu E, Miyazawa K (1986) Intersubject variability

in cardiac output-O2 uptake relation of men during exercise. J

Appl Physiol 61:2168–2174

66 Eur J Appl Physiol (2007) 101:61–66

123