ORIGINAL ARTICLE

Fabio Esposito Æ Franco M. Impellizzeri

Vittoria Margonato Æ Rosalba Vanni Æ Giuliano Pizzini

Arsenio Veicsteinas

Validity of heart rate as an indicator of aerobic demandduring soccer activities in amateur soccer players

Accepted: 23 June 2004 / Published online: 22 July 2004� Springer-Verlag 2004

Abstract In order to validate the use of heart rate (HR)in describing and monitoring physiological demandsduring soccer activities, the HR versus oxygen uptake( _V O2) relationship determined on the field during soc-cer-specific exercises was compared to that found in thelaboratory during treadmill exercise. Seven male ama-teur soccer players [mean (SE), age 25.3 (1.2) years,body mass 72.9 (2.1) kg, stature 1.76 (0.03) m] per-formed three trials on the field (two laps of a purpose-made circuit including a variety of soccer activities) atdifferent intensities (moderate, high and very high,according to their rate of perceived exertion) and anincremental test on a treadmill in the laboratory. HRincreased linearly with _V O2 during both field and labo-ratory tests according to exercise intensity (P<0.01).The mean correlation coefficients of the HR– _V O2 rela-tionships obtained in the laboratory and on the fieldwere 0.984 (0.012) and 0.991 (0.005) (P<0.001), re-spectively. The mean value of the HR– _V O

2regression

equation slope and intercept obtained in laboratory[0.030 (0.002) and 79.6 (4.6), respectively] were not sig-nificantly different compared to those found on the field[0.032 (0.003) and 76.7 (9.7)]. The present study seems toconfirm that HR measured during soccer exerciseseffectively reflects the metabolic expenditure of thisactivity. Thus, with the aid of laboratory reference tests,the physiological demands of soccer activities can becorrectly estimated from HR measured on the field inamateur soccer players.

Keywords Heart rate Æ Oxygen uptake Æ Soccer ÆPhysiological evaluation

Introduction

Soccer is one of the most popular sports in the worldand several studies investigated the physiological de-mands of soccer activities (Ekblom 1986; MacLarenet al. 1988; Rohde and Espersen 1988; Van Gool et al.1988; Tumilty 1993; Bangsbo 1994a, b; Felci et al. 1995;Reilly 1997). A widespread indirect method to evaluatethe metabolic demands during a soccer game or specifictraining is to record heart rate (HR) during socceractivities, which is aided by the use of comfortable andaccurate HR telemetric systems (Ali and Farrally 1991).However, a direct measurement of individual oxygenuptake ( _V O2) in the laboratory should always be per-formed in order to attempt to retrieve metabolic infor-mation from HR measured on the field. Indeed, basedon the linearity of the relationship between HR and _V O2

(Astrand and Rodahl 1977), the individual HR– _V O2

regression line obtained in the laboratory could then beused to determine the exercise intensity and the physi-ological demands for the specific HR measured on thefield. However, this methodology can be applied onlywhen the relationship determined in the laboratory hasbeen verified to be similar to that found on the fieldduring specific tasks at all exercise intensities.

Soccer is a very peculiar team sport, where continu-ous activities, such as walking and running, are ofteninterspersed with intermittent tasks, such as sprinting,jumping, kicking and dribbling. These factors, togetherwith thermal and emotional stresses typical of compet-itive soccer games, could introduce possible estimationerror sources and alter the linearity of the HR– _V O2

relationship on the field, especially at low levels of effort.Despite this indirect _V O2 estimation strategy is com-monly considered acceptable (Ekblom 1986; MacLarenet al. 1988; Rohde and Espersen 1988; Van Gool et al.1988; Bangsbo 1994a, Bangsbo b; Reilly 1997), Tumilty(1993) suggested that HR may not correctly reflect _V O2

during soccer activities. Moreover, Ogushi et al. (1993)measured _V O2 in two college soccer players during a

F. Esposito (&) Æ F. M. Impellizzeri Æ V. MargonatoR. Vanni Æ G. Pizzini Æ A. VeicsteinasInsitute of Physical Exercise, Health and Sports,University of Milan, Via Colombo 71,20133 Milan, ItalyE-mail: fabio.esposito@unimi.itTel.: +39-02-50314649Fax: +39-02-50314649

Eur J Appl Physiol (2004) 93: 167–172DOI 10.1007/s00421-004-1192-4

simulated soccer game and found that exercise intensityexpressed as _V O2 was �25% lower than that estimatedusing the HR– _V O2 relationship determined in the lab-oratory.

Some studies measured the exercise _V O2 using por-table gas analyzer systems in order to better understandthe physiological demands of various soccer-specificexercises in amateur (Kawakami et al. 1992), profes-sional (Hoff et al. 2002) and youth (Felci et al. 1995)soccer players. However, to our knowledge, the regres-sion line of the HR– _V O2 relationship obtained on thefield during soccer activities has never been compared tothat obtained in the laboratory.

Thus, the aim of the present study was to comparethe HR– _V O2 relationship determined on the field duringsoccer-specific tasks at different intensities to that foundin the laboratory during a standard treadmill test.Should the relationship be the same, the use of HRmonitoring on the field would be a valid indicator of thephysiological demands of the soccer-specific exercises inamateur players.

Methods

Subjects

Seven male amateur soccer players volunteered in thisstudy. The players trained with the team (sixth divisionof the Italian soccer league) twice a week, plus theweekly soccer game. Their characteristics are shown inTable 1. They were fully informed of the aims and theprocedures of the study and signed an informed consentform. The study was approved by the Ethical Committeeof the University of Milan. The experiments compliedwith the current laws of the country in which they wereperformed.

The subjects were instructed not to eat for at least 3 hbefore testing sessions. They were also asked to avoidbeverages containing caffeine for at least 8 h beforetesting, and intense exercise in the previous 48 h.

Tests were performed on the field and in the lab-oratory within 2 weeks. For each subject, field testsalways preceded the laboratory test in order to be ableto reproduce environmental conditions in the labora-tory similar to those found on the field. During bothfield and laboratory tests, expired respiratory gaseswere measured using a breath-by-breath automatedportable gas analysis system (K4 b2, Cosmed, Italy).Flow, volume and gas concentrations were calibratedbefore each test using standard procedures. The very

light (600 g) portable unit used in this investigation,which also provided HR measurements, has been re-ported to be reliable and valid for field measurementsof soccer exercises (Kawakami et al. 1992), as well asthose in the laboratory.

Field test

In order to reproduce the typical soccer activities(intermittent nature and movement patterns), a modifiedcircuit from Ekblom (1998) was used (Fig. 1). The cir-cuit was proposed to test soccer players and reproducethe most important soccer activities, such as jumping,changes in direction, walking, jogging, running (withand without ball), forward, backward and lateralmovements, etc. To this original circuit, some otherexercise modalities, such as passing and kicking, wereadded to include all the movement patterns and activi-ties commonly performed during soccer training orgame.

Field tests were performed outdoors, where averagetemperature ranged between 28 and 30�C, and relativehumidity between 60 and 70%. After about 10 min ofwarm up, consisting of jogging and running, astretching routine and a few high-intensity sprints, allsubjects underwent three trials (two times round thecircuit for each trial) at three different intensities:moderate, high and very high (maximal). To modulateintensity the subjects were asked to use their perceivedexertion quantified at the end of every intensity trialusing the Borg scale (0–10 point scale). In order toobtain correct feedback from the players, they werepreviously instructed in how to use the scale properly.Between each trial, athletes rested for 10 min. Duringthis phase, capillary blood samples (5 ll) were col-lected from the ear lobe and analysed using a portablesystem (Lactate Pro 1710, Arkray, Japan), at minutes1, 3 and 5.

Average data corresponding to the second lap of eachintensity trial were used to establish the individualaverage HR and _V O2 values on the field.

Laboratory test

The temperature of the laboratory was set at 28�C andrelative humidity at 60–70% in order to reproduce anambient condition similar to the outdoor field tests.After 10 min of warm up, the subjects performed anincremental exercise on a treadmill (770S, RAM, Padua,Italy) at a constant slope of 1%. Initial speed was set at9 km h)1 and was increased by 3 km h)1 after each step.Each workload lasted 5 min and a resting period of atleast 5 min was allowed between two subsequent steps.During the recovery phase, capillary blood samples(5 ll) were collected from the ear lobe at minutes 1, 3and 5, and analysed using the same portable system asthat used on the field. The maximal oxygen uptake

Table 1 Characteristics of the soccer players [mean (SE)]

Age (years) 25.3 (1.2)Stature (m) 1.76 (0.03)Body mass (kg) 72.9 (2.1)Body fat (%) 8.1 (2.2)Experience (years) 13.7 (2.3)

168

( _V O2max) criteria were: (1) a plateau in _V O2 (change<2.1 ml kg)1 min)1), (2) peak blood lactate concen-tration above 8 mmol l)1, or (3) a respiratory exchangeratio (R) above 1.15 (Duncan et al. 1997). Average datafrom the last minute of each step were used to determinethe individual HR and _V O2 values.

Statistical analysis

Data are presented as means (SE). Before using para-metric tests, the assumption of normality was verifiedusing the Kolmogorov–Smirnov test and Lillieforsprobabilities. For each subject, the linear regressions of_V O2 versus HR obtained in the laboratory and duringthe field test were determined. Group regression analyseswere not performed as this statistical approach couldmask the individual relationships (Swain and Leutholtz1997). For each regression the slope, intercept andPearson’s product-moment correlations were deter-mined. Paired t-tests were used to compare the slopesand intercepts of the HR– _V O2 individual regressionlines obtained in the two conditions (laboratory andfield). From the average HR data recorded during thesecond lap of every intensity field trial, the correspond-ing _V O2 was estimated using the individual HR– _V O2

regression equations obtained in laboratory. The esti-mated _V O2 was then compared to the measured _V O2

utilizing a two-way ANOVA for repeated measure-ments. When a statistical difference was detected, aBonferroni post-hoc test was applied. The level of sig-nificance was set at P<0.05.

Results

In Fig. 2, HR and _V O2 data from a representativesubject during the field test at three different exerciseintensities are shown. For our purposes, average datafrom the whole second lap were utilized.

The mean durations of the moderate, high and veryhigh-intensity trials were 411 (14), 326 (9) and 266 (4) s,respectively. During the field exercise at moderateintensity, average HR and minute ventilation ( _V E) were156 (4) beats min)1 and 68 (3) l min)1, respectively. The_V O2 reached at this level of exercise was 2.49(0.05) l min)1, and the corresponding carbon dioxideproduction ( _V CO2) was 2.50 (0.07) l min)1. At the endof this exercise, lactate concentration [La)] was 1.9(0.2) mmol l)1, with a mean rating of perceived exertion(RPE) of 2.3 (0.2).

At high intensity, HR and _V E were 173 (5) beats -min)1 and 96 (5) l min)1, respectively. Average _V O2 and_V CO2 were 3.08 (0.09) l min)1 and 3.31 (0.13) l min)1,respectively. [La)] at this intensity was 4.4 (0.5) mmoll)1, with a mean RPE of 4.4 (0.2).

Lastly, at very high-intensity exercise, metabolic andcardio-respiratory values are presented in Table 2.Briefly, HR reached average values of 186 (6) beats -min)1 and average _V O2 was 3.49 (0.29) l min)1. [La)]

Fig. 1 Soccer-specific circuit used in the present study (modifiedfrom Ekblom 1988). Arrows indicate the direction of the movementduring the exercise. Open circles The part of the track where ballwas involved (passes, kicks, dribbling, etc.), filled circles referencecones positioned on the field. Dashed lines The position where theintensity of the exercise slowed down (jogging or walking, at anintensity corresponding to the level of the exercise: moderate, highor very high)

169

reached values of 13.4 (0.1) mmol l)1, with a mean RPEof 9.1 (0.4).

During field tests at moderate, high and very highintensity, HR was 82 (2), 91 (1) and 97 (1)%, respec-tively, while _V O2 was 70 (2)%, 85 (2)% and 94 (2)%,respectively, of HR and _V O2 at maximal exercise in thelaboratory (see Fig. 3).

During the incremental maximal tests on the tread-mill, all the subjects met at least one of the _V O2max

criteria (Duncan et al. 1997). The set of data used tocalculate the regression line of the HR– _V O2 relationshipdetermined in the laboratory was averaged from the lastminute of each step. Maximal values reached during thelaboratory test were not statistically different from thosefound on the field, as shown in Table 2.

The two average regression lines obtained from theindividual HR– _V O2 regressions determined on the fieldand in the laboratory are shown in Fig. 4. The meanvalue of the HR– _V O2 regression line slope obtained inlaboratory [0.030 (0.002)] was not significantly differentcompared to that found on the field [0.032 (0.003)].Similarly, the mean value of the HR– _V O2 regressionline intercept obtained in the laboratory [79.6 (4.6)] wasnot significantly different compared to that found onthe field [76.7 (9.7)]. The mean correlation coefficientsof the HR– _V O2 obtained in laboratory and on the fieldwere 0.984 (0.012) and 0.991 (0.005) (P<0.001),respectively.

Discussion

The main finding of the present study is that the rela-tionship between HR and _V O2 determined in amateurplayers during soccer-specific exercises, simulating atraining circuit suggested in the literature (Ekblom1998), is not statistically different from that found in thelaboratory. Consequently, from the average HR re-corded during soccer activities, it seems to be possible toestimate the corresponding _V O2 by using the HR– _V O2

relationship determined in the laboratory.For this study, with the aim to reproduce all the

movement patterns and activities commonly performedduring soccer training or games, we modified the circuitproposed by Ekblom (1998) by adding walking, kickingand ball passes, which were not included in the originalcircuit. In this modified circuit, soccer players wereasked to perform three trials at different intensities(moderate, high and very high). Each trial was carriedout twice to reach a minimum of about 4 min of exerciseduration. Then, the individual HR– _V O2 regression linesfrom the field data were determined using the second lapaverage values, when HR and _V O2 data reached almostconstant average values (see Fig. 2). In this way, usingthe mean HR and _V O2 of the whole second lap, weincluded several factors supposed to affect the HR– _V O2

linearity (intermittent nature of the effort, sprint,jumping and static contractions).

Fig. 2 Example of raw datafrom a typical subject obtainedduring the field test. Grey areaThe range of time (second lap)from which heart rate (HR) andoxygen uptake ( _V O2) valueswere averaged to determine theHR– _V O2 relationship duringeach exercise intensity

Table 2 Maximal values relative to field (second lap of the veryhigh intensity trial) and laboratory tests. _V O2 Oxygen uptake,_V CO2 carbon dioxide production, _V E minute ventilation, R respi-ratory exchange ratio, RR respiratory rate, HR heart rate, [La) ]lactate concentration. Data are expressed as means (SE)

Field Laboratory

_V O2 (l min)1) 3.49 (0.11) 3.74 (0.17)_V O2 (ml kg)1 min)1) 48.1 (2.1) 51.7 (2.4)_V CO2 (l min)1) 4.14 (0.17) 4.18 (0.21)_V E (l min)1) 147.0 (14.5) 144.5 (5.8)R 1.19 (0.01) 1.12 (0.02)RR (breathsÆmin)1) 58 (2) 58 (2)HR (beatsÆmin)1) 186 (6) 191 (4)[La)] (mmol l)1) 13.4 (0.6) 13.5 (0.8)

Fig. 3 HR and _V O2 average values determined on the field at thedifferent exercise intensities as a percentage of the maximum HRand _V O2 values measured during the test on the treadmill

170

The average _V O2 values, measured on the field ateach exercise intensity, were not statistically differentfrom those estimated from the HR– _V O2 relationshipobtained in the laboratory (see Fig. 5). This findingconfirms the validity of HR monitoring to determine themetabolic demands during soccer activities, at least inthe range of the investigated intensities on the field [80–97% of laboratory heart rate maximum (HRmax) cor-responding to 70–94% of _V O2max]. These values aresimilar to those reported in the literature relative toactual games, when 80–90% of HRmax and 70–75% of_V O2max were measured (Ekblom 1986; Bangsbo 1994a,b; Felci et al. 1995).

The results of this investigation seem to be similar tothose found by Hoff et al. (2002) in six professionalsoccer players, when the average _V O2 values measuredduring a dribbling track and a small-group play (five vsfive) were close to the HR– _V O2 regression line deter-mined during a treadmill test. However, in the study of

Hoff et al. (2002), the incremental test protocol(1 km h)1 every minute) was probably too short tovalidate the _V O2 estimation from the HR recordedduring soccer activities, and to compare the HR– _V O2

relationship found on the field to that found in thelaboratory (Ekblom 1986; MacLaren et al. 1988; Rohdeand Espersen 1988; Van Gool et al. 1988; Bangsbo1994a, b). Indeed, constant work-loads of at least 4 min,as adopted in ours as well as in previous studies, and notan incremental protocol as in Hoff’s study (Hoff et al.2002), would be required in order to reach a steady-statecondition and to establish the HR– _V O2 relationship(Astrand and Rodahl 1977). Moreover, Hoff et al.(2002) concluded that HR monitoring is a valid measureof the exercise intensity during soccer-specific trainingsituations by plotting only one experimental point cal-culated from a specific maximal field test over the HR–_V O2 regression line obtained in the laboratory.

On the other hand, Tumilty (1993) suggested that HRmay not correctly reflect _V O2 during soccer activities, asintermittent or variable speed activities typical of soccermay elevate HR disproportionately to _V O2. Moreover,Ogushi et al. (1993) used the average HR and _V O2

measured on the field during a 90-min simulated friendlygame, and compared the results with the HR– _V O2

relationship determined in the laboratory. They found adifference of about 25% between estimated and mea-sured _V O2 and concluded that the discrepancy wasmainly due to the different _V O2 response during inter-mittent soccer activities compared to the steady-statelaboratory test. However, the use of Douglas bags forexpired gas collection could have limited the normalmovement of the players because of the weight (1.2 kg)and size of the equipment. Moreover, the lack of thelaboratory test protocol information, the small numberof subjects (only two players) and the short length of thegas collection time (2–3 min each half-time) make theinterpretation of their results difficult. Ogushi’s (Ogushiet al. 1993) conclusion is also not in agreement with thestudy of Drust et al. (2000), which showed similaraverage HR and _V O2 during intermittent exercise,reproducing a soccer game and continuous running atthe same average speed on a treadmill.

Among other factors that could influence the HR–_V O2 relationship, thermal and emotional stresses havebeen proposed, particularly during real competitivegames. Despite field tests being performed under hotand humid conditions, the laboratory temperature andhumidity were set accordingly in order to reproduce asimilar milieu. However, although field tests were car-ried out under occasionally sunny conditions, it ispossible that high environmental temperature influenceon HR was counterbalanced by the effect of wind andair exposure on body temperature regulation. Unfor-tunately, the design of the present study did not allowus to obtain more insights into the effects of emotionalstress on the HR– _V O2 relationship. Indeed, our resultsfrom the field were obtained during soccer-specificexercises, where stress plays a minimal role compared

Fig. 4 Scatter plot of the HR versus _V O2 data obtained from testsin the laboratory (dashed line) and on the field (continuous line). Theregression equations represented in the figure were calculated usingthe average slopes and intercepts of the individual regressions ofthe seven amateur soccer players

Fig. 5 Comparison between _V O2 estimated from the laboratorytest and _V O2 measured during the soccer circuit at different exerciseintensities (n=7). No significant interaction between the twomeasurement procedures (estimated vs measured) was found. ABonferroni post-hoc test revealed significant differences betweenmoderate versus high, high versus very high and moderate versusvery high

171

to real competitive games. It should be also taken intoaccount that stress can influence cardiovascularparameters at rest and during mild exercise, with neg-ligible effect at high and very high intensities of exer-cise. This aspect of the study, though, may requirefurther investigation.

Even anaerobic contribution should be taken intoaccount as a factor affecting metabolic demands. HR iscommonly a good indicator of aerobic metabolism, butit cannot reflect anaerobic contribution. In our investi-gation both peak [La)] and _V E (Table 2) reached duringthe last step in the laboratory were not statistically dif-ferent from those found at the end of the very highintensity trial on the field. This finding suggests thatanaerobic contribution was similar in both exercisemodalities. On the contrary, Drust et al. (2000) found asignificantly higher _V E during intermittent than duringcontinuous exercise at the same average speed. Theysuggested that the greater ventilation during intermittenteffort could be due to an increased metabolic acidosis asa consequence of a greater contribution of the anaerobicenergy system. The exercise intensity reached in theirstudy was related to the intensity observed during a realsoccer game, and resulted in about 68% of _V O2max. Inour study, the exercise intensity reached during the veryhigh-intensity field exercise reached almost the maximalvalue found in the laboratory [94 (2)% of _V O2max), wellabove the average intensity used by Drust et al. 2000.Moreover, the different exercise modalities, such asjumping, kicking and running used in the present study,can justify the discrepancy. Lastly, the average durationof Drust’s study (Drust et al. 2000) was about 46 min,separated into two halves of 23 min each. In our study,the average duration of maximal exercise on the fieldwas of about 4.5 min. For these reasons, it seems diffi-cult to compare the two studies as far as anaerobicmetabolism is concerned.

In conclusion, the similarity of the HR– _V O2 rela-tionship between field soccer-specific exercises and lab-oratory tests on a treadmill found in the present studyconfirms that average HR during soccer activities effec-tively reflects aerobic metabolic demands. Thus, thephysiological load can be determined using a laboratoryreference test. Our study underlines the appropriate useof HR for correct and useful soccer training strategies, atleast in amateur soccer players. Moreover, our findingsmight be relevant also to other sports or activities, wheresimilar intermittent phases of the exercise are inter-spersed with continuous aerobic performances. How-ever, a greater number of subjects, including professionalplayers with higher maximum aerobic power, would beopportune in order to extend the results of the presentwork to a wider population. We hope that by adding

experimental evidence on the importance of HR moni-toring during specific activities coupled to laboratorytests, trainers and coaches will be encouraged to use amore scientific approach to sports.

Acknowledgements The authors wish to acknowledge Emiliano C’eand Marcello Iaia for their technical assistance in data collectionduring the experiments. We also thank the soccer players involvedin this study for their committed participation.

References

Ali A, Farrally M (1991) Recording soccer players’ heart ratesduring matches. J Sports Sci 9:183–189

Astrand PO, Rodahl K (eds) (1977) Textbook of work physiology.McGraw-Hill, New York

Bangsbo J (1994a) Energy demands in competitive soccer. J SportsSci 12[Spec No.]:S5–12

Bangsbo J (1994b) The physiology of soccer—with special refer-ence to intense intermittent exercise. Acta Physiol Scand619[Suppl]:1–155

Drust B, Reilly T, Cable NT (2000) Physiological responses tolaboratory-based soccer-specific intermittent and continuousexercise. J Sports Sci 18:885–892

Duncan GE, Howley ET, Johnson BN (1997) Applicability of_V O2max criteria: discontinuous versus continuous protocols.Med Sci Sports Exerc 29:273–278

Ekblom B (1986) Applied physiology of soccer. Sports Med 3:50–60

Ekblom B (1998) A field test for soccer players. Sci Football 1:13–15

Felci U, De Vito G, Macaluso A, Marchettoni P, Sproviero E(1995) Functional evaluation of soccer players during child-hood. Med Sport (Roma) 48:221–235

Hoff J, Wisloff U, Engen LC, Kemi OJ, Helgerud J (2002) Soccerspecific aerobic endurance training. Br J Sports Med 36:218–221

Kawakami Y, Nozaki D, Matsuo A, Fukunaga T (1992) Reliabilityof measurement of oxygen uptake by a portable telemetricsystem. Eur J Appl Physiol 65:409–414

MacLaren D, Davids K, Isokawa M (1988) Physiological strain ina 4-a-side soccer. In: Reilly T, Lees A, Davids K, Murphy WJ(eds) Science and football. Spon, London, pp 76–80

Ogushi T, Ohashi J, Nagahama H, Isokawa M, Suzuki S (1993)Work intensity during soccer match-play (a case study). In:Reilly T, Clarys J, Stribbe A (eds) Science and football. Spon,London, pp 121–123

Reilly T (1997) Energetics of high-intensity exercise (soccer) withparticular reference to fatigue. J Sports Sci 15:257–263

Rohde HC, Espersen T (1988) Work intensity during soccertraining and match-play. In: Reilly T, Lees A, Davids K,Murphy WJ (eds) Science and football. Spon, London, pp 68–75

Swain DP, Leutholtz BC (1997) Heart rate reserve is equivalent to% _V O2 reserve, not to % _V O2max. Med Sci Sports Exerc 29:410–414

Tumilty D (1993) Physiological characteristics of elite soccerplayers. Sports Med 16:80–96

Van Gool D, Van Gerven D, Boutmans J (1988) The physiologicalload imposed on soccer players during real match-play. In:Reilly T, Lees A, Davids K, Murphy WJ (eds) Science andfootball. Spon, London, pp 51–59

172