Development of muscle strength in relation to traininglevel and testosterone in young male soccer players

L. HANSEN,1 J. BANGSBO,1 J. TWISK,2 AND K. KLAUSEN11Department of Human Physiology, Institute of Exercise and Sport Sciences, Universityof Copenhagen, DK-2100 Copenhagen, Denmark; and 2Institute for Research in ExtramuralMedicine, Vrije Universiteit, 1081 Amsterdam, The Netherlands

Hansen, L., J. Bangsbo, J. Twisk, and K. Klausen.Development of muscle strength in relation to training leveland testosterone in young male soccer players. J. Appl.Physiol. 87(3): 11411147, 1999.Isometric and functionalstrength of ninety-eight 11-yr-old male soccer players at anelite (E) and nonelite (NE) level were determined (34 times)through a 2-yr period, and the changes were related to growthand maturation. The initial isometric strength for extensionwith dominant leg [1,502 6 35 (E) vs. 1,309 6 39 (NE) N],extension with nondominant leg (1,438 6 37 vs. 1,267 6 45N), extension with both legs (2,113 6 76 vs. 1,915 6 72 N),back muscles (487 6 11 vs. 414 6 10 N), abdominal muscles(320 6 9 vs. 294 6 8 N), and handgrip (304 6 10 vs. 259 6 8 N)increased by 1540% during the period. Broad jump in-creased (P , 0.05) by 15 (E) and 10% (NE). The E players hadhigher (P , 0.05) initial isometric strength and broad jumpperformance compared with NE players, and differences weremaintained throughout the period (multiple ANOVA for re-peated measures) also when adjustment was made for age,dimensions, testosterone, and insulin-like growth factor I(generalized estimating equations analyses). The develop-ment of strength for both E and NE players together wassignificantly (P , 0.001) related to changes in serum testoster-one concentrations. The present data indicate that testoster-one is important for development of strength in young boysand that, independent of serum testosterone concentration, Eplayers have developed greater muscle strength comparedwith NE players.

elite and nonelite players; age; dimensions

IN ADULTS, substantial knowledge is present aboutfactors that determine muscle strength and its changewith training (e.g. Refs. 13, 15), but less information isavailable about development of muscle strength inchildren. Rochcongar et al. (22) found that youngFrench elite soccer players had greater isokinetic legstrength compared with high school students, indicat-ing that soccer training has an effect on the develop-ment of muscle strength. In contrast, Maffulli et al. (18)found that athletic boys (including soccer players) untilthe age of 15 yr had similar isometric quadricepsstrength as did nonathletic boys, and after this age thestrength of the athletic group was significantly highercompared with nonathletic boys. The latter finding mayindicate that the training responses are affected bymaturation.

Muscular strength increases more or less linearlywith age from early childhood in boys. Strength isknown to be related to the physiological cross-sectionalarea of the muscle and hence, according to a dimen-sional analysis, related to the second power of bodyheight. During growth the cross-sectional area wouldthen be expected to increase, with the square of theincrease in the linear dimension. Some studies haveshown that strength development in boys improvesmore than can be explained by increase in heightsquared (1, 8), indicating that factors other than quan-titative changes play a role in the development ofstrength. Thus the marked acceleration of musclestrength during puberty observed in boys is possiblyrelated to the elevated levels of circulating androgenhormones in adolescents. In a cross-sectional study of11- to 13-yr-old athletic boys, Mero et al. (20) found thatmuscle fiber area correlated well with serum testoster-one. Maturation of the metabolic response to exercisemight be related to the hormonal changes [increases intestosterone, estradiol, growth hormone, and insulin-like growth factor I (IGF-I)] occurring during puberty(9, 19). On the basis of a difference in increase instrength between boys and girls, Parker et al. (21)suggested that testosterone may stimulate musclegrowth. The growth-promoting effect of growth hor-mone is mediated by somatomedins, particularly IGF-I(16). However, no longitudinal study has measuredchanges in blood hormone concentrations and relatedthem to changes in muscle size and strength of childrenand adolescents (23).

As indicated by Rochcongar et al. (22), soccer trainingat an elite level might increase leg strength, but it isalso possible that the boys selected for the elite level arestronger because of higher levels of circulating hor-mones.

The aim of this investigation was to study thedevelopment in strength of boys playing soccer at anelite and a nonelite level and to examine the associationbetween the development in strength and testosteroneconcentration. Both questions were addressed withcorrection for age, body size dimensions, and IGF-I.

METHODS

Subjects. One hundred and ten young male soccer playersfrom seven successful clubs in the area of Copenhagen, at thehighest level in their age category, were recruited as subjects.All participants and their parents gave their informed con-sent, and the study was approved by the Ethics Committee ofCopenhagen, Denmark (KF 01132/95). Clubs that had atleast four teams in the same age categories were selected (toensure differences between elite and nonelite players). The

The costs of publication of this article were defrayed in part by thepayment of page charges. The article must therefore be herebymarked advertisement in accordance with 18 U.S.C. Section 1734solely to indicate this fact.

8750-7587/99 $5.00 Copyright r 1999 the American Physiological Society 1141http://www.jap.org

boys were included in the study at the age of 1012 yraccording to the selection age at the competition levels.One-half of the boys were recruited from the best team towhich they were selected by the coach (elite players), and theother one-half were recruited from the lowest ranked team(nonelite players) from the same club. Measurements weretaken three times at 0.5-yr intervals for all the boys, and, inaddition, 28 of the subjects (16 elite, 12 nonelite) were alsostudied a fourth time. Eight boys were excluded from thestudy because they were not members of the same teamduring the whole study, two because they did not want toparticipate after the first test and two because they moved toanother part of the country. Thus 98 subjects were includedwith 48 boys in the elite group and 50 boys in the nonelitegroup. During the study some boys stopped playing soccerand were then excluded from subsequent tests. Thus 87 boyswere tested three times, and 28 of these were also tested afourth time.

The players included in the study had participated inorganized soccer for 6.3 (elite) and 4.4 (nonelite) yr with asignificant difference (1.9 yr; P , 0.05) between the groups. Asshown in Table 1, the elite players were playing soccer formore hours per week and were in general more physicallyactive compared with the nonelite players. The leisure timeactivity registered consisted mostly of soccer with friends butalso roller-skating and high-activity play as well as participa-tion in other organized sports.

The age of the subjects was assessed to the nearest 0.01 yr.Standing and sitting height were measured by a stadiometerto the nearest 0.1 cm, and body weight was measured to thenearest 0.1 kg by using a spring balance. The body mass index(BMI) was calculated as body weight (kg) divided by height(m) squared. Bicipital, tricipital, subscapular, and suprailia-cal skinfolds were measured with a Harpenden skinfoldcaliper, and the sum of these four skinfolds was calculated.

The pubertal developmental stages were recorded by oneexperienced pediatric endocrinologist on the basis of assess-ment of secondary sex characteristics by using the criteria ofTanner (25) and from testicular volume estimated frommeasurements of the size of the testes by using a Praderorchidometer (29). Blood samples were drawn from an antecu-bital vein between 1600 and 1730 and were centrifuged.Serum was stored at 220C and later analyzed for levels oftestosterone and IGF-I. The sensitivity of the assay for IGF-Iwas 0.041 g/l (12), and the sensitivity for testosterone was0.23 nmol/l. Values less than assay sensitivity were assignedthe value of assay sensitivity.

Strength measurements. The subjects were all familarizedwith the testing procedures as well as with the investigatorsbefore the test. All subjects had a standardized warm-upperiod, including 5 min of cycling on a Monarck bike, beforethe strength measurements. All subjects started with thebroad jump followed by measurements of isometric strength.Broad jump was performed as a two-foot takeoff and landing.The takeoff was from behind a line on the floor, and the

landing was on a 2-cm-thick mat on which the subjects wereinstructed to land on their feet. The distance from the takeoffto the point where the nearest heel touched the mat wasmeasured, and the best of three recorded trials was used asthe performance score (cm). The maximal voluntary isometricstrength [maximal voluntary contraction (MVC)] of the legextensors was measured by using a strain-gauge dynamom-eter in a standardized seated position with support of theback (4). To measure MVC of the trunk muscles in a standingposition, a strain-gauge dynamometer was connected to aframe placed around the trunk 20 cm below the shoulders bythe use of two straps (4). Grip strength was measured with ahand dynamometer in subjects while seated and for thedominant arm only. The boys were all encouraged to thehighest effort by the investigators, and the best of threeattempts was accepted as maximal.

The training regimen and competition intensity were evalu-ated for a subgroup of the original subjects (n 5 30, 10 teams).The competition intensity was evaluated from heart ratemeasurements by a heart rate monitor (Polar) both duringcompetition and while the subjects ran on a treadmill withsimultaneous measurements of oxygen uptake. The eliteplayers had higher relative oxygen uptake in competitioncompared with the nonelite group (79.4 6 5.3 vs. 67.3 6 10%;P , 0.05). For the elite and nonelite players, 27 vs. 7% of thetraining consisted of fitness training (sprint run, etc.), 61 vs.39% was technical training, and 12 vs. 54% was play. Nosupplemental weight training was used.

Statistics. To assess the longitudinal relationship betweenstrength and soccer ability (elite or nonelite), two analyseswere carried out. 1) In the first analysis, the differences inlongitudinal development of strength measurements betweenelite and nonelite soccer players were analyzed with multipleANOVA (MANOVA) for repeated measures (SPSS; Ref. 24). 2)In the second analysis, the longitudinal relationship betweenstrength and soccer ability (elite or nonelite) was analyzedwith generalized estimating equations (GEEs) (30), a longitu-dinal linear regression technique that is extensively de-scribed elsewhere (26, 27). The advantages of using thismethod are that all available longitudinal data are used toestimate the regression coefficients and that the method issuitable for designs with unequally spaced time intervals.Furthermore, it allows a correction for both time-dependentand time-independent covariates and the method takes intoaccount that the repeated observations on each individual arenot independent. GEE analysis was also carried out toanalyze the longitudinal relationship between strength pa-rameters and testosterone concentration. For all GEE analy-sis, a univariate analysis was carried out first. After that,separate analyses were carried out correcting for age, bodysize dimensions, and IGF-I concentration. All GEE analyseswere carried out with the Statistical Package for InteractiveData Analysis (10). Significance was accepted at P # 0.05.

Table 1. Amount of time with organized soccer training (including competitions) and with physicalactivity in leisure time, including participation in other sports

Test Round

1 2 3 4

Organized Leisure Organized Leisure Organized Leisure Organized Leisure

Elite 6.1* 11.7* (19) 6.2* 10.7* (16) 6.5* 11.9 (19) 7.4* 11.4 (9)Nonelite 5.1 6.4 (26) 5.2 7.0 (20) 4.8 10.6 (9) 5 10.2 (10)

Values are given in h/wk with % in parentheses. *Significant different from nonelite players, P , 0.05.

1142 STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS

RESULTS

The characteristics of the subjects are presented inTable 2. The elite players were slightly older than thenonelite players (0.4 yr; P , 0.05). When adjustmentwas made for age, the elite players were significantlytaller (P , 0.05) and had lower values for skinfoldmeasurements (P , 0.05). The elite players had greatertesticular volume than did the nonelite players andhigher serum testosterone concentration (Fig. 1; P ,0.05). No significant differences between the groups inBMI or IGF-I (Fig. 2) were present.

Figures 36 show the development in strength param-eters for elite and nonelite players. Differences indevelopment between the two groups were analyzed byMANOVA for repeated measures. The results of theseanalyses (Table 3) showed that elite players had highervalues (P , 0.05) compared with nonelite players for allstrength parameters throughout the measurement pe-riod. The increase in strength over time was, however,similar in both groups, i.e., no significant elite/nonelite-time interaction for any of the strength parameters. No

significant differences in development between thegroups according to Tanner stages were present(MANOVA for repeated measures).

The results from the GEE analysis regarding thelongitudinal relationship between being an elite player/nonelite player and strength development are pre-sented in Table 4. In univariate analysis a significantpositive relationship was found between all strengthparameters and being an elite player. In general, withan adjustment for age, the regression coefficients forelite/nonelite players decreased slightly; a more markeddecrease was found for the relationship with leg exten-sion by using both legs. Adjustment for body dimen-sions also led to a decrease in regression coefficients.When the relationships between elite/nonelite playerand strength were adjusted for height, weight, and sumof skinfolds, as well as for numbers of years of trainingin organized soccer, only the relationships with backmuscles and handgrip remained significant (P , 0.05).When adjustment was made for serum testosteroneand IGF-I, a small decrease in regression coefficientswas observed; however, the positive relationships be-tween all strength parameters and being an elite playerremained highly significant.

Table 2. Characteristics of subjects

Test Round

1 2 3 4

Elite Nonelite Elite Nonelite Elite Nonelite Elite Nonelite

Age, yr 11.960.5 11.660.7 12.460.5 12.160.7 12.960.5 12.560.7 13.560.6 13.860.4n 48 50 44 47 44 43 16 12Height, cm 152.766.8 147.466.6 155.767.3 150.166.9 16068.3 154.368.1 166.3610.0 160.467.2Body mass, kg 41.065.9 37.966.5 43.665.97 40.067.0 46.667.1 43.067.7 53.268.8 47.767.3BMI, kg/m2 17.561.6 17.462.1 17.961.5 17.762.3 18.161.6 17.862.1 19.162.1 18.562.7Skinfold, mm 27.666.9 33.7614.3 28.366.8 35.1615.4 27.566.4 36.1615.6 29.267.3 39.3618.5Testes volume, ml 5.863.5 3.961.9 7.664.9 5.063.3 9.365.4 6.663.9 12.166.5 9.365.4n 45 49 43 47 43 42 16 12

Values are means 6 SD; n, no. of subjects. BMI, body mass index; skinfold, sum of 4 skinfolds (biceps, triceps, subscapularis, suprailiac). Forheight, body mass, BMI, and skinfold n is the same as for age. Each test round was carried out at 0.5-yr intervals.

Fig. 1. Serum testosterone concentrations. Measurements were takenat 0.5-yr intervals starting when the subjects had a mean age of 11.96 0.5 (SD) (elite) and 11.6 6 0.7 yr (nonelite). Values are means 6 SE;n, no. of subjects. Dotted line indicates that the last test roundincluded a reduced number of players (n 5 28). Difference (adjustedfor age) between groups is significant for 4 test rounds (P 5 0.015)with a tendency for significance present for 3 test rounds (P 5 0.076).

Fig. 2. Insulin-like growth factor I levels in elite and nonelite youngmale soccer players. Values are means 6 SE; n, no. of subjects. Nosignificant difference between elite and nonelite players was present.

1143STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS

The results of the GEE analysis regarding the longi-tudinal relationship between strength and serum testos-terone are presented in Table 5. Univariate analysisshowed a significant positive relationship between de-velopment in all strength parameters and serum testos-terone concentration. Adjustment for age, body dimen-sions, and IGF-I showed more or less the same pictureas for the longitudinal relationships between being anelite/nonelite player and strength, i.e., in general, adecrease in regression coefficients. The adjustment forbody weight, height, and sum of skinfolds led to adramatic decrease in regression coefficients; i.e., noneof the relationships between serum testosterone andstrength parameters was significant.

DISCUSSION

The present study showed that both elite and nonel-ite players through a 2-yr period had an increase instrength that was related to the levels of serum testos-terone, indicating that testosterone is important fordevelopment of muscle strength in young boys. Inaddition, a strong positive relationship between beingan elite player and the level of strength was observed.This relationship was independent of testosterone andIGF-I, indicating that the greater strength was notsolely due the level of serum testosterone.

Increases in muscle strength with age in young boyscannot simply be explained by growth, because it hasbeen shown in both prepubertal and pubertal boys thatstrength increases more rapidly than does height (6). It

Fig. 3. Isometric strength for abdominal and back muscles measuredin standing elite and nonelite young male soccer players. Values aremeans 6 SE. Measurements were taken at 0.5-yr intervals startingwhen the subjects had a mean age of 11.9 6 0.5 (SD) (elite) and 11.6 60.7 yr (nonelite). Values are means 6 SE; n, no. subjects for eliteplayers/nonelite players in each test round. Difference between thegroups is significant: abdominal, P 5 0.011; and back, P 5 0.001.

Fig. 4. Isometric handgrip strength (dominant hand) in elite andnonelite young male soccer players. Values are means 6 SE; n, no. ofsubjects. Differences between the groups are significant, P 5 0.002.

Fig. 5. Isometric strength for leg extensors measured in seated elite(E) and nonelite (N-E) young male soccer players. Values are means 6SE; n, no. of subjects for elite players/nonelite players in each testround. Differences between the groups are significant: both legs, P 50.004; dominant leg, P 5 0.002; and nondominant leg, P 5 0.01.

Fig. 6. Broad jump performance in elite and nonelite young malesoccer players. Values are means 6 SE; n, no. of subjects. Differencebetween the groups is significant, P 5 0.002.

1144 STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS

is more likely to be due to an interrelationship betweenseveral factors such as age, stature, weight, musclesize, and maturation (endocrine and neurological). Fromexperimental data and from the recognition that testos-terone has a prominent anabolic effect, it has beensuggested that testosterone is responsible for the in-

crease in strength in male individuals at puberty (5). Inthe present study this is confirmed by a significantpositive relationship between development in allstrength parameters and serum testosterone concentra-tion. This relationship is dependent on anatomic dimen-sions and skinfold thickness, indicating that thesefactors also play a role in the development of strength.Asmussen and Heebll-Nielsen (2) suggested that, be-sides dimensions, age per se has a positive influence onmuscular strength, especially in tests that require ahigh degree of neuromuscular coordination. We foundthat the relationship between strength developmentand changes in serum testosterone was independent ofage, except for leg extension with both legs. It has beenshown for both children and adults (14, 28) that MVCfor leg extension with both legs are less than the sum ofMVC for each leg, indicating a limit in neural output.Furthermore, in the present study the percentile differ-ence between the sum of the strength of each leg andboth legs decreased with age [from 39 to 15% (elite) andfrom 34 to 22% (nonelite)], which is in agreement withAsmussen and Heebll-Nielsen (3), who showed a

Table 3. Results of MANOVA for 3 test rounds, carriedout at 0.5-yr intervals on isometric strength and broadjump, testing the effects of team (elite/nonelite), timeby team, and time of measurement

TeamTime by

TeamTime of

Measurement

Handgrip 0.002 0.594 ,0.001Abdomen 0.011 0.052 ,0.001Back 0.001 0.371 ,0.001Both legs 0.004 0.191 ,0.001Nondominant leg 0.010 0.762 ,0.001Dominant leg 0.002 0.999 ,0.001Broad jump 0.002 0.696 ,0.001

Values are P values adjusted for age (age as covariate). MANOVA,multiple ANOVA for repeated measures.

Table 4. Longitudinal relationships between being an elite or a nonelite player and development of isometricstrength/broad jump, adjusted for confounding factors

UnivariateAnalysis

AbdominalMuscles

BackMuscles Handgrip

Leg Extension

BroadJump

Dominantleg

Nondominantleg

Bothlegs

Elite/noneliteP 0.000 0.000 0.000 0.000 0.000 0.001 0.000b 244.1 271.3 249 2239.3 2242.8 2413.4 213.395% CI 267.320.8 2101.3 to 241.3 272.8 to 225.2 354.3 to 2124.2 2365.4 to 2120.2 2651.2 to 2175.6 219 to 27.7

Adjusted for ageP 0.048 0.000 0.000 0.057 0.046 0.226 0.001b 222.1 254.1 241.9 2112.8 2121.7 2165.4 29.895% CI 244 to 20.17 281.0 to 227.3 263.4 to 220.3 2228.93.2 2241.4 to 22.1 2433.5102.6 215.8 to 23.8

Adjusted fordimensions:

Height and bodymass

P 0.214 0.000 0.000 0.014 0.029 0.073 0.058b 211.6 246.8 230.7 2100.6 293.8 2180 26.195% CI 229.96.7 267.2 to 226.3 247.8 to 213.6 2181.2 to 220.1 2177.8 to 29.8 2376.5 to 21.45 212.50.2

Height, bodymass14skinfolds

P 0.502 0.000 0.005 0.133 0.316 0.428 0.507b 26.2 231.4 225.3 267.5 244.9 279.4 21.995% CI 224.4 to 211.9 247.6 to 215.1 243 to 27.6 2155.520.6 2132.542.7 2275.7117 27.43.7

Adjusted forserum tes-tosterone

P 0.006 0.000 0.000 0.004 0.008 0.032 0.001b 229.4 262.2 239.7 2158.1 2152.0 2249.8 29.995% CI 250.4 to 28.5 289 to 235.4 261.1 to 218.3 2266.5 to 249.9 2264.4 to 245.8 2478.5 to 221.1 215.7 to 24.1

Adjusted for IGF-IP 0.003 0.000 0.000 0.002 0.003 0.014 0.000b 235.4 266.8 246 2185.6 2178.8 2305.7 211.595% CI 259.1 to 211.7 294.9 to 238.7 269.2 to 222.8 2301.7 to 269.4 2298.1 to 259.6 2550 to 261.4 217.4 to 25.5

Adjusted for yearsof training

P 0.91 0.038 0.004 0.763 0.97 0.719 0.159b 21.6 234.4 234.3 22.2 2.9 254.7 24.995% CI 29.4 to 229.4 266.9 to 21.9 257.8 to 210.7 2122.2166.4 2145.4151.1 2353.4243.9 211.81.9

Values are P values, longitudinal regression coefficients (b), and 95% confidence intervals (CI) obtained from generalized estimatingequations analysis. IGF-I, insulin-like growth factor I.

1145STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS

gradual decrease of this difference in male subjectsaged 1535 yr. Thus dependency of age when examin-ing extension with two legs (with some degree ofneuromuscular coordination) could be explained by arequirement for a neuromuscular maturation possiblyrelated to age.

The nonelite players, while not as strong as the eliteplayers, had almost the same strength values as didDanish schoolboys, aged 11 yr, who were examined in1981 (11). It seems likely that the development of legmuscle strength in particular would give an advantageto the elite soccer player. Leatt et al. (17) showed agreater isokinetic and explosive strength in Canadiannational soccer players who were under 18 yr oldcompared with the national players who were under 16yr old. The elite players in the present study were alsostronger compared with the nonelite players when acorrection was made for the small difference in agebetween the two groups. The elite players were tallerand more mature compared with the nonelite players,so the increase could be due to growth, maturation, orcompetition level. To examine this relationship, theGEE analysis was carried out. The difference in strengthbetween elite players and nonelite players appears notto be due to the difference in height between the twogroups because the difference was independent of di-mensions, except for leg extension with both legs andbroad jump, which only revealed a tendency for indepen-dence (P 5 0.073 and P 5 0.058, respectively). Therelationship between strength development and elite/nonelite was independent of serum testosterone andIGF-I, indicating that the development in strength wasrelated to factors associated with being an elite player

independent of testosterone concentration. The reasonfor this increase in strength may be due to a greaterrelative increase in muscle mass of the elite players andthus a larger cross-sectional area of the muscles. Alter-natively, it may have been caused by qualitative changesin the muscles such as a lower ratio of connective tissueto muscle tissue so that the same mass of musculaturemay be brought to produce more tension in the eliteplayers. Leatt et al. found that elite players had morelean body mass compared with normal subjects. In thepresent study no difference in BMI was found betweenthe two groups, but the elite players had less subcutane-ous fat evaluated from skinfold measurements, whichindicate a larger lean body mass in the elite players,possibly caused by muscle hypertrophy as a response totraining. When the sum of skinfolds was included withdimensions in the GEE analysis of relationships be-tween elite/nonelite and strength development, a depen-dency was found for leg extensions, abdominal muscles,and broad jump but not for back muscles and handgripstrength. This indicates that the development instrength is related to some extent to a hypertrophy ofthe muscles. It is also plausible that part of thedifference in strength may result from a better masteryof the neuromuscular system in the elite players causedby the training regimen that the elite players wereexposed to from an early age. The fact that the eliteplayers were initially stronger compared with thenonelite players could partly explain why the increasein strength for the elite players during the test perioddid not lead to an increase in the difference between thetwo groups because strength is known to increase morefrom lower initial levels (e.g., Ref. 7). It cannot, how-

Table 5. Longitudinal relationships, for both elite and nonelite players as one group, between serum testosteronelevels and development of isometric strength/broad jump, adjusted for confounding factors

UnivariateAnalysis

AbdominalMuscles

BackMuscles Handgrip

Leg Extension

BroadJump

Dominantleg

Nondominantleg

Bothlegs

TestosteroneP 0.000 0.000 0.000 0.000 0.000 0.000 0.000b 11 11.2 6.4 56.6 60 99.1 2.395% CI 7.914.1 7.714.8 3.19.8 36.776.4 37.982.1 50.7147.6 1.72.9

Adjusted for ageP 0.011 0.001 0.012 0.055 0.029 0.442 0.000b 3.1 5.7 4.1 13 17.9 14.4 1.195% CI 0.75.6 2.49.0 0.97.4 20.326.2 1.833.9 222.351 0.51.7

Adjusted for dimen-sions

Height and bodymass

P 0.829 0.165 0.575 0.484 0.572 0.636 0.029b 0.3 2.3 0.9 5.0 5.2 9.6 0.895% CI 22.32.8 20.95.5 22.24.1 29.119.2 212.723.0 230.149.3 0.11.4

Height, bodymass14 skin-folds

P 0.604 0.726 0.790 0.991 0.834 0.758 0.466b 20.7 20.7 0.5 0.1 21.9 25.8 0.395% CI 23.31.9 3.25.1 21.12 214.113.9 219.215.4 242.831.1 20.41

Adjusted for IGF-IP 0.000 0.000 0.000 0.000 0.000 0.008 0.000b 8.8 8.9 6.1 46.3 46.7 86.9 2.295% CI 5.212.4 4.613.2 2.99.3 22.770.3 20.972.6 22.6151.2 1.52.9

1146 STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS

ever, be excluded that the differences between the eliteand nonelite players are due to an early selection ofboys with higher strength for the elite group.

In summary, development of isometric strength andperformance in broad jump was related to changes inserum testosterone concentrations but also influencedby body size, indicating that testosterone is importantfor development of strength in young boys. Further-more, elite players were stronger than nonelite playersindependent of testosterone concentration also withcorrection for body size, indicating that being an eliteplayer per se affected the development of strength.

The authors gratefully acknowledge Dr. Med. Jrn Muller fordoing the measurements of testicular volume and the Tanner stageevaluations and further for the valuable discussions regarding theinterpretations of the maturational data.

This study was supported by grants from the Danish Association ofSoccer, the Danish Association of Sports, and Team Danmark.

Address for reprint requests and other correspondence: L. Hansen,Institute of Exercise and Sport Sciences, Dept. of Human Physiology,Universitetsparken 13, DK-2100, Copenhagen, Denmark(E-mail: L1Hansen@aki.ku.dk).

Received 9 April 1998; accepted in final form 26 April 1999.

REFERENCES

1. Asmussen, E., and K. Heebll-Nielsen. A dimensional analy-sis of physical performance and growth in boys. J. Appl. Physiol.7: 593603, 1955.

2. Asmussen, E., and K. Heebll-Nielsen. Physical performenceand growth in children. Influence of sex, age and intelligence. J.Appl. Physiol. 8: 371380, 1956.

3. Asmussen, E., and K. Heebll-Nielsen. Isometric musclestrength of adult men and women. In: Communications From theTesting and Observation Institute. Copenhagen, Denmark: Dan-ish Natl. Assoc. of Infantile Paralysis, 1961, no. 11, p. 313.

4. Asmussen, E., K. Heebll-Nielsen, and S. Molbech. Methodsfor evaluation of muscle strength. In: Communications From theTesting and Observation Institute. Copenhagen, Denmark: Dan-ish Natl. Assoc. for Infantile Paralysis, 1959, no. 5, p. 144.

5. August, G. P., M. M. Grumbach, and S. L. Kaplan. Hormonalchanges in puberty. III. Correlations of plasma testosterone, LH,FSH, testicular size and bone age with male pubertal develop-ment. J. Clin. Endocrinol. Metab. 34: 319326, 1972.

6. Blimkie, C. J. R. Age- and sex-associated variation in strengthduring childhood: anthropometric, morphologic, neurologic, bio-chemical, endocrinologic, genetic, and physical activity corre-lates. In: Perspectives in Exercise and Sports Medicine, edited byC. V. Gisolfi and D. R. Lamb. Carmel, IN: Benchmark, 1989.

7. Blimkie, C. J. R. Resistance training during pre- and earlypuberty: efficacy, trainability, mechanisms, and persistence.Can. J. Sport Sci. 17: 264279, 1992.

8. Carron, A. V., and D. A. Bailey. Strength development in boysfrom 10 through 16 years. Monogr. Soc. Res. Child Dev. 39: 137,1974.

9. Cooper, D. M. Cardiorespiratory and metabolic responses toexercise: maturation and growth. In: The Child and the Adoles-cent Athlete, edited by O. Bar-Or. Champaign, IL: BlackwellScientific, 1996.

10. Gebski, V., O. Leung, D. McNeil, and D. Lunn. SPIDA UserManual. New South Wales, Australia: Macquarie Univ., 1992,ver. 6.

11. Heebll-Nielsen, K. Muscle strength of boys and girls, 1981compared to 1956. Scand. J. Sports Sci. 4: 3743, 1982.

12. Juul, A., P. Bang, N. T. Hertel, K. Main, P. Dalgaard, K.Jrgensen, K, J. Muller, K. Hall, and N. E. Skakkebk.Serum insulin-like growth factor-1 in 1030 healthy children,adolescent, and adults: relation to age, sex, stage of puberty,testicular size, and body mass index. J. Clin. Endocrinol. Metab.78: 744752, 1994.

13. Klausen, K. Strength, and weight-training. In: Physiology ofSports, edited by T. Reilly. London: Spon, 1990, p. 4167.

14. Klausen, K., B. Schibye, and B. Rasmussen. A longitudinalstudy of changes in physical performance of 1015-year-old girlsand boys. Children and Exercise XIII, edited by S. Oseid andK.-H.Carlsen. Champaign, IL: Human Kinetics Books, 1989, p.113122.

15. Komi, P. V. (Editor). Strength and Power in Sport. The Encyclo-pedia of Sports Medicine. Oxford, UK: Blackwell Scientific, 1992,vol. 3.

16. Kulin, H. E., and J. Muller. The biological aspects of puberty.Pediatr. Rev. 17: 7582, 1996.

17. Leatt, P., R. J. Shepard, and M. Plyley. Specific musculardevelopment in under-18 soccer players. J. Sports Sci. 5: 165175, 1987.

18. Maffulli, N., J. B. Kings, and P. Helms. Training in elite youngathletes [the training of young athletes (TOYA) study]: injuries,flexibility and isometric strength. Br. J. Sp. Med 28: 123126,1994.

19. Marshall, W. A., and J. M. Tanner. Puberty. In: HumanGrowth, edited by F. Falkner and J. M. Tanner. New York:Plenum, 1986, vol. 2.

20. Mero, A., L. Jaakkola, and P. V. Komi. Relationships betweenmuscle fibre characteristis and physical performance capacity intrained athletic boys. J. Sports Sci. 9: 161171, 1991.

21. Parker, D. F., J. M. Round, P. Sacco, and D. A. Jones. Across-sectional survey of upper and lower limb strength in boysand girls during childhood and adolescence. Ann. Hum. Biol. 17:199211, 1990.

22. Rochcongar, P., R. Morvan, J. Jan, J. Dassonville, and J.Beillot. Isokinetic investigation of knee extensors and kneeflexors in young French soccer players. Int. J. Sports Med. 9:448450, 1988.

23. Rowland, T. W. Developmental Exercise Physiology. Cham-paign, IL: Human Kinetics, 1996.

24. SPSS. SPSS-X, Users Guide (3rd ed.). Chicago, IL: SPPS, 1988.25. Tanner, J. M., and R. H. Whitehouse. Clinical longitudinal

standards for height, weight, height velocity, weight velocity, andstages of puberty. Arch. Dis. Child. 51: 170179, 1976.

26. Twisk, J. W. R., H. C. G. Kemper, G. J. Mellenbergh, and W.van Mechelen. Relation between the longitudinal developmentof lipoprotein levels and lifestyle parameters during adolescenceand young adulthood. Ann. Epidemiol. 6: 246256, 1996.

27. Twisk, J. W. R., H. C. G. Kemper, G. J. Mellenbergh, and W.van Mechelen. Relation between the longitudinal developmentof lipoprotein levels and biological parameters during adoles-cence and young adulthood in Amsterdam. J. Epidemiol. Commu-nity Health 50: 505511, 1996.

28. Vandervoort, A. A., D. G. Sale, and J. Moroz. Comparison ofmotor unit activation during unilateral and bilateral leg exten-sion. J. Appl. Physiol. 56: 4651, 1984.

29. Zachmann, M., A. Prader, H. P. Kind, H. Haflinger, and H.Budliger. Testicular volume during adolescence. Helv. Paediatr.Acta 29: 6172, 1974.

30. Zeger, S. L., and K.-Y. Liang. Longitudinal data analysisfor discrete and continuous outcomes. Biometrics 42: 121130,1986.

1147STRENGTH AND DEVELOPMENT IN YOUNG SOCCER PLAYERS