STUDY

Prediction of race walking © by IAAF

23:4; 35-41, 2008

performance via laboratory

and field tests
By Andrew Drake, Robert James

ABSTRACT AUTHORS
It may be useful to be able to predict race Andrew Drake, PhD, is the Director of the
walking performance based on relation- Sport & Exercise Science Applied
ships between standard race velocities Research Group in the Department of
and physiological variables such as blood Biomolecular & Sport Sciences at
lactate (Blac) or maximum oxygen uptake Coventry University in Great Britain. He
(VO2max), however laboratory tests for is a UK Athletics Level IV Performance
this purpose are impractical in most situ- Coach in Race Walking and a Level III
ations. The aim of this study was to devel- Performance Coach in Middle & Long
op a method to predict competition per- Distance Running.
formance and/or physiological variables Robert James, PhD, is the Associate Head
from performances in other distances of the Department of Biomolecular &
and/or field based tests such as a time Sport Sciences at Coventry University, in
trial. Laboratory treadmill tests were con- Great Britain.
ducted on 68 athletes to find Blac, VO2max
and race walking economy (ml O2/kg/km).
Twenty-one of the same subjects then
completed a 2000m full-effort time
trial for calculation of their race veloci-
Introduction
ty (km/h). The values obtained were
compared and combined with the per-
revious studies of race walking
formances of the athletes over standard
race walking distances and the per-
formances of elite athletes to produce a
P (DRAKE et al. 2003; HAGBERG &
COYLE, 1983; YOSHIDA et al.
1989) have concluded that success in this
nomogram (a diagram representing the
discipline is probably related to the ability to
relations between three or more vari-
attain and sustain a high race walking veloci-
able quantities) that could be used to
ty without accumulation of blood lactate (Blac).
predict any of the variables if one or
Therefore, it may be useful for coaches, ath-
more of the values was known. The find-
letes and others to be able to predict perform-
ings support the use of time trials to
ance based on the relationships between
predict velocity at VO2max and show
standard race velocities and Blac or other
there is a link between laboratory and
physiological variables such as maximum
performance variables that can be used
oxygen uptake (VO2max). Although it is often
in coaching.
impractical to perform laboratory tests for this
purpose, it could be efficacious to predict

New Studies in Athletics • no. 4/2008 35

Prediction of race walking performance via laboratory and field tests

race walking performance and/or physiologi- had Coventry University’s ethical approval.
cal variables from performances in other dis- The mean age, height, body mass and maxi-
tances and/or a field based test such as a mum oxygen uptake was 27 ± 8 years; 1.73
time trial. ± 0.08m; 64.3 ± 9.8kg; and 62.9 ± 11.3
ml/kg/min respectively.
For example MERCIER et al. (1986) devel-
oped a nomogram to predict performance Laboratory test: treadmill protocol
equivalence for distance runners; and BILLAT The athletes completed between six and
et al. (1994) and BERTHON et al. (1997) nine four-minute stages of race walking on
developed field based tests to calculate a motorised treadmill. All tests began at a
velocity at maximum oxygen uptake (v- 1% gradient, increasing by 0.5km/h each
VO2max). DABONVILLE et al. (2003) found stage, with a starting speed 2.0km/h below
that the five-minute running test of BERTHON the current race speed for 10km of the sub-
et al. (1997) was reliable for estimating v- ject. On completion of each stage a 20µl
VO2max from only one trial. BILLAT et al. arterialised capillary blood sample was
(1994) found that the duration of running per- obtained from the ear lobe for the determi-
formance that could be maintained at v- nation of Blac values used to identify lactate
VO2max was 5:21 ± 1:23. BERTHON et al. turn point (LTP). Expired air was collected
(1997) found running v-VO2max correlated into a Douglas bag for the last 60 seconds
best with performance over 3000m, which of each stage to determine oxygen uptake
varied from 8:08 to 15:36 (r = 0.97, p <0.05, (VO 2) and race walking economy (ml
n = 9 male runners); and DANIELS (1998) O2/kg/km). When heart rate (HR) exceeded
described v-VO2max as representative of the 95% of the predicted maximum or Blac
speed of a running race lasting between 10 exceeded 4mmol/l the treadmill gradient
and 15 minutes, therefore a 2000m race walk was increased by 1% every 60 seconds.
would compare time wise with a 3000m run The test continued until volitional exhaus-
performance. Moreover laboratory-, perform- tion for the determination of maximum oxy-
ance- and field-based variables could be gen uptake (VO2max). The velocity at lac-
combined to construct the interrelationships tate turn point (v-LTP) was the race walking
between them, e.g. to create a nomogram to speed at which there was an abrupt and
predict VO2max, v-VO2max, 2000m time trial exponential increase in Blac values. The
performance and 3000m, 5000m, 10km, velocity at VO 2max (v-VO 2max) was
20km and 50km race walk performance if one resolved by linear regression on sub-maxi-
or more of the values was known. mal race walking speed and VO2 values.

The purpose of this study was to develop a Field test: 2000m time trial
nomogram (a diagram representing the rela- Thirteen male and eight female race
tions between three or more variable quanti- walkers from the main subject group
ties) to predict performance equivalence for (above) also race walked an all out 2000m
race walking, which would enable coaches time trial on a 400m Mondo surface athlet-
and athletes to predict competition perform- ics track for calculation of v-2km (km/h).
ance and/or physiological variables from per- The time trial was undertaken within 24
formances in other distances and/or a field hours of completing the discontinuous
based tests such as a time trial. incremental treadmill test described above.
Subjects were informed they should race
Methods walk as fast as possible and to treat the
time trial as a race effort. Time trials began
Subjects after subjects had completed a 20 minute
Forty-five male and twenty-three female warm-up. Tests were undertaken on wind-
race walkers participated in this study, which less days when the track was dry. Subjects

36 New Studies in Athletics • no. 4/2008

 Prediction of race walking performance via laboratory and field tests

completed their time trials alone on the Results

track to avoid creating a competitive envi-
ronment. The mean ± SD age, height, body 2000m time trial
mass and VO2max of the 2000m time trial The 2000m time trial performance time
group was 22 ± 9 years; 1.75 ± 0.07m; was 554 ± 65s. The v-2km was compared to
62.3 ± 9.1kg; and 55.6 ± 8.9ml/kg/min v-VO2max and race walk competition per-
respectively. formances over 3000m, 5000m, 10km,
20km and 50km recorded within four weeks
Construction of interrelationships between pre- or post-laboratory test. The v-2km (13.2
variables ± 1.6km/h) was 0.2% higher than v-VO2max
The value for v-2km was compared to v- (13.1 ± 1.5km/h) (p > 0.05); and 10.9% high-
VO2max identified in the laboratory and race er than v-LTP (11.7 ± 0.7km/h) (p < 0.05).
walk performances over 3000m, 5000m, The v-VO2max accounted for 94% of the
10km, 20km and 50km by the athletes taking variance in 2000m time trial performance
part in the present studies. Moreover, race when analysed by multiple stepwise linear
walk performances over 3000m, 5000m, regression. The relationship between
10km, 20km and 50km attained by athletes v-VO 2max and v-2km was resolved by
ranked in the world, British or Italian top 50 following the “field-laboratory” equation in
between 1999 and 2003 were combined to Table 1 (R2 = 0.96, n = 21). To test whether
create a nomogram to predict VO2max, v- the distribution of the VO2max, v-VO2max
VO2max, 2000m time trial performance and and v-2km data measured were significantly
3000m, 5000m, 10km, 20km and 50km race different from the normal distribution a Kol-
walk performance if one or more of the values mogorov-Smirnov test (D) was used (FIELD,
was known. 2005). The distribution of the VO2max val-

Table 1: Relationships between paired race distances analysed using linear regression,
where 3000m, 5000m, 10km, 20km and 50km = hh:mm:ss (Predictive equations were
established from World, British and Italian top 50 ranked performances 1999 – 2003, which
were not significantly different (p >0.05) from the present study. #linear regression not per-
formed due to n = 2.)

Field-laboratory Common race Performance-field

walking economy

v-2 km (km/h) = VO2 (ml/kg/min) = 5.2482 v-3 km (km/h) =

1.1042 v-VO2max (km/h) – 1.4011 treadmill speed (km/h) – 12.334 0.8624 v-2 km (km/h) + 1.6626

r = 0.96*, R2 = 0.96, n = 21 r = 0.85*, R2 = 0.996, n = 68 r = 0.93*, R2 = 0.76, n = 21

Paired race distances (h:mm:ss) used to construct nomogram

3000m - 5000m 5000m - 10 km 10km - 20km 20km - 50km
5000m = 1.4468 10 km = 2.0906 20 km = 2.1031 50 km = 2.8868
3000m + 0.0024 5000m – 0.0002 10 km – 0.0009 20 km – 0.0039
R2 = 0.90, n = 31 R2 = 0.92, n = 44 R2 = 0.87, n = 44 R2 = 0.74, n = 63
Paired race distances (h:mm:ss) from present study for comparison
3000m - 5000m 5000m - 10 km 10km - 20km 20km – 50km
5000m = 1.463 10 km = 1.8573 20km = 2.2811
3000m + 0.0025 5000m + 0.0034 10km - 0.0064 n = 2#
R2 = 0.82, n = 30 R2 = 0.87, n = 34 R2 = 0.86, n = 28

New Studies in Athletics • no. 4/2008 37

Prediction of race walking performance via laboratory and field tests

ues, (D(21) = 0.12 (p > 0.05)), v-VO2max Discussion

values (D(21) = 0.21 (p > 0.05)) and v-2km
values (D(21) = 0.16 (p > 0.05)) indicated a Certain assumptions were inherent in the
normal distribution. construct of the nomogram, e.g. the relation-
ship between race walking speed (km/h) and
Interrelationships between variables VO2 max (ml/kg/min). The regression equation
Data from the present study was combined created a curve that assumes common race
to resolve the relationship between race walk- walking economy; however the common
ing speed and VO2, i.e. “common race walk- curve ignores the differences in economy and
ing economy” (Table 1, R2 = 0.996, n = 68); the differences in VO2max. In the race walking
and the “performance-field” based relation- nomogram an athlete completing a 2000m
ship between v-2km and v-3km was resolved time trial in 8:00 would be predicted as having
from race performances completed by the a v-VO2max of 15.0km/h and a VO2max of
athletes who completed the 2000m time trial about 66.0ml/kg/min, however in reality a v-
(Table 1, R2 = 0.76, n = 21). The model con- VO2max of 15.0km/h may be attained by an
struct was completed from analysis of 182 athlete with a lower VO2max but superior race
paired race walk times of athletes over the walking economy or by an athlete with a high-
3000m, 5000m, 10km, 20km and 50km dis- er VO2max but inferior race walking economy,
tances (Table 1). The construct from the inter- i.e. a race walker using the nomogram to pre-
relationship between laboratory-, perform- dict VO2max may find a different value pre-
ance- and field-based variables was the dicted to that which could be measured com-
nomogram (Figure 1) that can be used to pre- pleting the discontinuous incremental tread-
dict performance in race walking events and mill race walking protocol.
VO2max.
MERCIER et al. (1986) developed a nomo-
Present study v ranking data gram to predict performance equivalence for
The linear regression equations from Table distance runners and proposed a number of
1 predict small differences in performance. uses for such a tool, which are also applica-
For example, the VO2max of the male athletes ble to the nomogram presented here. Race
who completed a 2000m time trial was 55.6 walk performance may be predicted by inter-
± 8.9ml/kg/min with a corresponding v- polation or extrapolation, e.g. an athlete com-
VO2max of 13.2 ± 1.6km/h (Table 2), equating pleting a 2000m time trial in 8:00 s and 10km
to v-2km of 13.2km/h (“field-laboratory” time trial in 44:06 would be predicted to race
equation, Table 1) that is used to compute walk 5000m in 21:14; the performances of an
3000m performance from 2000m time (“per- athlete completing a 2000m time trial in 7:21
formance-field” equation, Table 1), i.e. v- and a 10km time trial in 41:04 could be
3000m of 13.02km/h. The linear regression extrapolated to predict to 50km in 3:59:58. A
equations from Table 1 predict a 1.6 % differ- further use for the nomogram could be to
ence between 5000m performances (22:26 determine the prerequisites to achieve a cer-
derived from ranking data versus 22:47 tain level of performance, such as a qualifying
derived from present study); a 1.3% differ- time, e.g. the “A” standard qualifying time for
ence between 10km performances (46:37 the Men’s 20km at the 2008 Olympic Games
derived from ranking data versus 47:13 was 1:23:00: an athlete seeking to attain this
derived from present study); and a 1.8% dif- level of performance would have had to be
ference between 20km performances able to complete time trials at 2000m in 7:08
(1:36:44 derived from ranking data versus or 3000m in 10:57, 5000m in 19:18 or 10km
1:38:30 derived from present study). The rela- in 40:03.
tionships between paired performances from
the present study compared to the ranking In previous studies (DRAKE et al. 2003;
data are shown in Figure 2. HAGBERG & COYLE, 1983; YOSHIDA et al.

38 New Studies in Athletics • no. 4/2008

 Prediction of race walking performance via laboratory and field tests

Figure 1: Nomogram to predict race walking performance at distances from 2000m to 50km
(VO2max is predicted by passing a horizontal line through the 2000m time trial (TT) performance.
The line that describes the race walking performance of an athlete at two distances allows pre-
diction of performance at a third distance.)

New Studies in Athletics • no. 4/2008 39

Prediction of race walking performance via laboratory and field tests

Figure 2: Relationships between paired performances from the present study compared to the
ranking data used (a. 3000m – 5000m; b. 5000m – 10km; c. 10km – 20km; and d. 20km –
50km) (Performances by athletes ranked in the World, British or Italian top 50 between 1999 and
2003 were used to construct the model (n = 56 male, 37 female) – www.iaaf.org; www.ukath-
letics.org.uk; and www.fidal.it).

1989) the relationships between several Conclusions

physiological variables and athlete perform-
ance suggested that success in race walk The 2000m time trial provided information on
competition was largely related to aerobic v-VO2max: v-2km and v-VO2max were not
endurance, i.e. the ability to attain and sus- statistically significantly different from each other
tain a high race walking velocity for a long (p >0.05) and r = 0.98 (p <0.05). Furthermore
period of time. However the present study the 2000m time trial and nomogram could be
did not establish the validity of the laborato- used for VO2max evaluation and to identify
ry based methods compared to the field prerequisite levels of performance required to
based 2000m time trial to determine the rel- achieve goals such as attaining championship
ative importance of the different variables in qualifying times. Even so, the 2000m time trial
performance prediction, e.g. laboratory warrants validation with more subjects to
based threshold values versus time trial per- strengthen the interrelationships with laborato-
formance. ry- and performance-based variables.

40 New Studies in Athletics • no. 4/2008

 Prediction of race walking performance via laboratory and field tests

Recommendations race walking nomogram, which is an easy to

use coaching tool for athletes and coaches.
The present study supports the use of a
field based 2000m race walking time trial as
a predictor of velocity at maximum oxygen Please send all correspondence to:
uptake in athletes competing in race walking Dr Andrew Drake
events. Moreover, the 2000m time trial pro- email: a.drake@coventry.ac.uk
vides a useful link between laboratory and Dr Robert James
performance variables in the construct of the e-mail: r.james@coventry.ac.uk

REFERENCES
BERTHON, P.; FELLMAN, N.; BEDU, M.; BEAUNE, B.; walkers (abstract). Journal of Sports Sciences, 23,
DABONNEVILLE, M.; COUDERT, J. & CHAMOUX, A. 199-200.
(1997). A 5-min running field test as a measurement of
maximal aerobic velocity. European Journal of Applied DRAKE, A.; WOOLEY, R.; JAMES, R.; COX, V.; GODFREY,
Physiology, 75 233-23. R. & BROOKS, S. (2007) Prediction of velocity at maximum
oxygen uptake in race walkers. Proceedings of the 12th
BILLAT, V.; RENOUX, J.C.; PINOTEAU, J. & KORAL- Annual Congress of the European College of Sport Sci-
SZTEIN, J. P. (1994). Times to exhaustion at 100% of ence, Jyväskylä, Finland.
velocity at VO2max and modelling of the time limit/velocity
relationship in elite long-distance runners. European Jour- FIELD, A. (2005). Discovering Statistics Using SPSS. 2nd
nal of Applied Physiology, 69 271-3. edition. London: Sage.

DABONNEVILLE, M.; BERTHON, P.; VASLIN, P. & FELL- HAGBERG, J. M. & COYLEe, E. F. (1983). "Physiological
MAN, N. (2003). The 5 min running field test: test and determinants of endurance performance as studied in
retest reliability on trained men and women. European competitive race walkers.” Medicine and Science in Sports
Journal of Applied Physiology, 88 353-360. and Exercise, 15 (4) 287-289.
DANIELS, J. T. (1998). Daniels' Running Formula. Cham- MERCIER, D.; LEGER, L. & DESJARDINS, M. (1986).
paign, IL: Human Kinetics. "Nomogram to predict performance equivalence in dis-
DRAKE, A.; COX, V.; GODFREY, R. & BROOKS, S. (2003). tance runners.” Track Technique, 94 3004-3009.
Physiological variables related to 20 km race walk per-
YOSHIDA, T.; UDO, M.; IWAI, K.; MURAOKA, I.; TAMA-
formance (abstract). Journal of Sports Sciences. 21,
KI, K.; YAMAGUCHI, T., & CHIDA, M. (1989) "Physiolog-
269–270.
ical determinants of race walking performance in female
DRAKE, A.; COX, V.; GODFREY, R. & BROOKS, S. race walkers.” British Journal of Sports Medicine, 23 (4),
(2005). Race walking economy of highly trained race 250-254.

New Studies in Athletics • no. 4/2008 41