Anthropometrics and electromyography as predictors for maximal voluntary isometric arm strength

Lara A.Green,David A.Gabriel*

Department of Kinesiology,Brock University,St.Catharines,ON L2S 3A1,Canada

Anthropometrics and electromyography as predictors for maximal voluntary isometric arm strength

Lara A.Green,David A.Gabriel*

Department of Kinesiology,Brock University,St.Catharines,ON L2S 3A1,Canada

Background:Muscularstrength can be conceptually determined by two components:muscle activation and size.Muscle activation by the central nervous system can be measured by surface electromyography(sEMG).Muscular size reflects the amountof contractile protein within a skeletal muscle and can be estimated by anthropometric measurements.The purpose of this study was to determine the relative contributions of size parameters and muscle activation to the prediction of maximal voluntary isometric elbow flexion strength.

Methods:A series ofanthropometric measurements were taken from 96 participants.Torque and root-mean-square(RMS)of the sEMG from the biceps brachiiwere averaged across three maximalvoluntary isometric contractions.A multiple linear regression analysis was performed based on a Pearson’s correlation matrix.

Results:Body weight(BW)accounted for 39.1%and 27.3%in males and females,respectively,and was the strongestpredictor of strength for males.Forearm length(L3)was the strongest predictor of strength in females(partial R2=0.391).Elbow circumference(ELB)accounted for a significant(p＜0.05)amountof variance in males butnotfemales.The addition of sEMG RMS as a third variable accounted for an average of 10.1%of the variance excluding the equation of BWand L3 in females.The strongestprediction equation included BW,L3,and ELB accounting for 55.6%and 58.5%of the variance in males and females,respectively.

Conclusion:Anthropometrics provide a strong prediction equation for the estimation of isometric elbow flexion strength.Muscle activation,as measured by sEMG activity,accounted for a significant(p＜0.05)amountof variance in mostprediction equations,however,its contribution was comparable to an additional anthropometric variable.

Copyright?2012,Shanghai University of Sport.Production and hosting by Elsevier B.V.All rights reserved.

Biceps brachii;Biomechanics;Gender differences;Multiple linear regression;Muscle force

1.Introduction

Muscular strength can be determined by two components: muscle activation and muscle size.The first of these twocomponents,muscle activation,is the result of efferent output from the central nervous system(CNS).1This includes the controlof motor unitrecruitment(the number of active motor units)and motor unit firing rate(the rate at which they fire). Motor unit recruitment and firing rate are reflected in the amplitude of the interference pattern of the summated action potentials recorded by surface electromyography(sEMG).2The second component of strength is based on the amount of contractile proteins within skeletalmuscle.3—5The amount of contractile tissue can be measured by cross-sectional area (CSA)and anthropometric measures used to infer muscle size.4,6

Itiswidely known thatCSA isatleastmoderately correlated (r=0.5—0.7)with voluntary strength regardless ofgender,age and training status.5,7The relationship between muscle size and force is of sufficient magnitude that the“specific tension”of a muscle iscommonly used in musculoskeletal modeling studies to predict force.8The specific tension of a muscle is the force normalized with respectto its CSA.

Krolland colleagues9extended the research in this field by developing strength prediction equations using non-invasive, simple measures of body weight(BW),body volume, segmentallimb lengths and volumes of the upper limb for both males and females.Multiple regression analysis revealed that the bestpredictorofelbow flexion strength was BWformales (R=0.69),and total upper limb volume for females (R=0.72).Kroll and colleagues9also determined that limb girths and lengths predictelbow flexion strength as wellas,or better than,segmental limb volumes thereby simplifying the methodology in this area.

Given the relationship between muscle activation(sEMG) and force10,11it would seem logical to add this variable to a multiple regression equation thatpredicts force.An equation that incorporates both anthropometric data and sEMG measurementshould theoretically capture the two components of muscle strength(size and muscle activation)and decrease the standard error of estimate.The presentstudy willtherefore determine the relative contributions of body size and muscle activation in a strength prediction equation.The hypothesis of this study is that adding muscle activation(sEMG)to anthropometrics willimprove the strength prediction equation.

2.Materials and methods

Ninety-six(46 males and 50 females),right-handed college age participants took part in the present study.Each subject was verbally acquainted with the experimental design and provided written,informed consent(REB#02-284).

2.1.Anthropometrics

Since this paper attempted to extend the work of Krolland colleagues9by adding muscle activation(sEMG),we collected the same anthropometric measurements used in that paper. Anthropometric data(Fig.1)were collected priorto the testing procedure by an experienced person who took the average of three measurements using a tape measure and the following landmarks:

Lengths

L1:acromion process to deltoid tubercle

L2:deltoid tubercle to olecranon process

L3:olecranon process to styloid process of the ulna

L4:styloid process of the ulna to tip of the third finger

Circumferences

AC:circumference at acromion process

DEL:circumference atdeltoid tubercle

ELB:circumference atolecranon process

Fig.1.Illustration of anthropometric length and circumference measurements collected.L1,acromion process to deltoid tubercle;L2,deltoid tubercle to olecranon process;L3,olecranon process to ulnar styloid;L4,ulnar styloid to tip of third finger;AC,girth at acromion process;DEL,girth at deltoid tubercle;ELB,girth atolecranon process;WJS,girth atstyloid process(wrist joint space);HND,girth at base of hand.

WJS:circumference atdistalspace to styloid process of the ulna

HND:thickness of the base of the hand,cross-sectional height of thenar and hypothenar eminence.

2.2.Set up and procedure

Testing took place within a Faraday cage,and was completed in one session.Participants were seated in an adjustable chair and fastened with Velcro?straps to reduce movement.The rightarm of the participantwas positioned in the sagittal plane,with the shoulder and elbow flexed to 90°within a jig designed to isolate the upper limb.With the wrist in neutralposition,a cuff was fastened proximalto the styloid process and attached to a load cell(JR3 Inc.,Woodland,CA, USA)to record force.Participants were asked to perform three 5-s MVCs separated by 3-min rest intervals.An oscilloscope (VC-6525;Hitachi,Woodbury,NY,USA)displaying the participant’s force trace was placed in front of the participant for visual feedback.Surface EMG was recorded while participants performed the contractions.

Each participant’s arm was shaved,abraded and cleansed with alcohol to reduce signal impedance to below 10 kΩ (Grass EZM5;Astro-Med Inc.,West Warwick,RI,USA). Motor points were determined using low level surface stimulation to elicit a visible twitch.Silver—silver chloride electrodes(Grass F-E9;Astro-Med Inc.)were then placed on the skin surface in line with the biceps brachiimuscle fibers,2 cm away from the motor point toward the distal tendon.The electrode configuration was bipolar with an inter-electrode distance of 2 cm.A 5-cm ground electrode(CF5000;Axelgaard Manufacturing Company Ltd.,Fallbrook,CA,USA) was placed on the participant’s clavicle.Surface EMG activity was amplifi ed 1000 times before being band-passed filtered between 3 and 1000 Hz(Grass P511,Astro-Med Inc.).

Allsignals were sampled at2048 Hz using a 16-bitanalogto-digital converter(NI PCI-6052E;National Instruments, Austin,TX,USA)controlled by a computer-based data acquisition system DASYLab(DASYTEC,National Instruments, Amherst,NH,USA).The data were collected and stored foroffline processing on a desktop computer(Seanix Technology Inc., Blaine,WA,USA).A one-second window centered at the middle of each contraction was used to extractthe mean force and root-mean-square(RMS)amplitude of the sEMG signal. The data used for analysis was the mean of three trials.

2.3.Statistics

The correlational approach followed was that described by Kroll and colleagues.9All statistical analysis was performed using SYSTAT(Systat Inc.,Evanston,IL,USA).First, a correlation(Pearson)matrix was constructed to determine which anthropometric measurements correlated highly with torque.A stepwise multiple linear regression analysis was performed based on the correlation matrix.In the first stage, the variable with the highest correlation,BW,was entered as the only predictor of strength.In the second stage,a second anthropometric variable was added.To avoid multicolinearity, only one of each length and circumference were chosen to be included in the primary equations.Forearm length(L3)was selected because it was highly correlated with torque for both males and females,and it is a measure of the lever length during elbow flexion.Elbow circumference(ELB)was selected because itwas highly correlated with torque for both males and females,and includes the size of the elbow flexor muscles at the joint crossing.Once the equation for BW and L3 or ELB was determined,sEMG RMS was added to the equation to determine the contribution of muscle activation. The predictive value of three anthropometric variables was also assessed.As well,prediction equations were performed using the fourlength measurements with the addition ofsEMG RMS,and the five circumference measurements with the addition of sEMG RMS,to determine the contribution of sEMG to each group of variables.

For each equation,the R2and partial R2were calculated to determine the strength of the equation and the relative contribution of the added variable,respectively.The standard error of the estimate(SEE)was calculated to help determine the benefit of adding another variable versus the cost of decreasing the degrees offreedom associated with the specific equation.Finally,an F-ratio was calculated to determine if there was a significant(p＜0.05)increase in the variance accounted-for by an additional variable,relative to the benchmark equation.12

3.Results

The mean±SD values for torque,sEMG RMS and anthropometric measurements are presented in Table 1.The results of the correlation matrix are presented in Table 2 and multiple linear regression analyses are presented in Table 3, for males and females,respectively.

3.1.Contribution of anthropometrics

The initialprediction equation with only BWaccounted for 39.1%and 27.3%of variance in elbow flexion strength in males and females,respectively(Fig.2).BWwas the strongest strength predictor for males.The addition of L3 to the equation improved strength prediction for both males and females. Based on the partial R2,L3 was the strongest strength predictor for females accounting for 39.1%of the variance. The addition of ELB to the initialequation with BWimproved the strength prediction for males with a significant(p＜0.05) partial R2of 12.5%;however,it had little effect on the equation for females.The best prediction equation for both males and females consisted of three anthropometric measures(BW, L3,and ELB),accounting for 55.6%and 58.5%of the total variance in strength,respectively(Fig.3).To compare lengths versus circumferences,overall prediction equations of all four lengths and all five circumferences were performed.In males, the circumferences were much stronger predictors compared to the lengths(R2=0.545 and 0.293,respectively).This wasnot the case in females where the lengths and circumferences equations were equal(R2=0.389).

Table 1Mean±SD values for subject physical characteristics,anthropometric measurements and sEMG root-mean-square(RMS).

Table 2Correlation matrix of anthropometric variables including body weight(BW),biceps brachiiroot-mean-square(RMS)and elbow flexion force.

Table 3Multiple regression analysis for maximal voluntary isometric force prediction in male and female.

3.2.Contribution of muscle activation(sEMG)

Muscle activation,as measured by sEMG RMS,was added to each of the equations containing two anthropometric variables(Fig.3).The addition of sEMG RMS to a prediction equation with BW and L3 resulted in a non-significant (p＞0.05)increase in variance-accounted-forin elbow flexion strength.The partial R2for males was 7.9%while itwas only 3.3%forfemales.The addition ofsEMG RMS to a prediction equation with BWand ELB resulted in a significant(p＜0.05) increase in the variance-accounted-for in elbow flexion strength,with a partial R2of 11.5%for males and 10.9%for females(Fig.2).The addition ofsEMGRMS to the fourlengths and to the five circumferenceswasfound to be statistically nonsignificant(p＞0.05)for both equations.The prediction equations and their results are detailed further in Table 3.

4.Discussion

4.1.Contribution of anthropometrics

In agreement with Kroll et al.,9BWalone was a moderate predictor of strength.BWis the most common anthropometric measure used in strength prediction3and wasused asthe basisof this regression analysis.The inclusion of a second anthropometric measure was determined based on both the correlation with strength and its biomechanical significance to elbow flexion.L3 was selected because it functioned as the biomechanical lever during the task.Forearm length was calculated from theolecranon process(joint)to thestyloid process(location of load cell).Since the elbow was fixed at 90°of flexion the distance from thejointto theload cellrepresentstheleverarm,or resistance arm of the movement.13The other anthropometric variable included in thesecond stageoftheregression waselbow circumference.ELB represents regionalmuscle mass due to its widely accepted high,positive correlation with force.5,7,14,15

Fig.2.Graphical results of the multiple linear regression analysis to predict maximal voluntary isometric elbow flexion force in males(A)and females (B).The total of each bar displays the total amount of variation predicted by the equation.The gray portion of each bar displays the additionalamountof partial variation predicted with the inclusion of the last variable in the equation.Variables in the prediction equations include:body weight(BW),length from olecranon process to styloid process(L3),circumference at olecranon process(ELB),and biceps brachii activity(RMS).

Fig.3.Graphical results of the total variance accounted for by prediction equations including three variables for males(black)and females(gray). Variables in the prediction equations include:body weight(BW),length from olecranon process to styloid process(L3),circumference at olecranon process (ELB),and biceps brachiiactivity(RMS).

Upper arm circumference is a popular measure used in force prediction equations for upper body exercises due to its high correlation(r=0.65—0.77)with force.7,14,16Likewise,in the presentstudy,ELB accounted for an additional12.5%and 18.9%of the variance in male elbow flexion strength,when added to equations of one and two variables,respectively (Table 3).The elbow circumference measure was,however, not statistically significant(p＞0.05)when added to either equation in females.Gender differences in strength and CSA are well-known and are more apparent for the upper versus lower limbs.17Although the amountofforce produced perunit CSA has been found to be equalbetween males and females,it cannotnecessarily be applied to circumference measurements. Miller et al.18and Kanehisa et al.19have found that females have an increased proportion of fat mass compared to lean tissue mass(muscle and bone).Therefore,circumference measurements may not be as representative of force per CSA in females as in males,and it ultimately was not as good of a strength predictor for females.

Scanlan and colleagues20performed a factor analysis on anthropometric variables used in a bench press prediction equation and was able to account for 68.4%of the total variance in force.The first factor consisted of body mass, muscle circumferences,and skinfolds accounting for47.8%of the variance in force.In contrast,the third factor included heightand limb lengths and accounted for only 7%of the total variance in force.The regression analysis for males in the presentstudy is consistentwith the findings of Scanlan etal.20because a circumference measure(ELB)had a greater impact on the equation for predicting elbow flexion strength than a length measure(L3).In contrast,the inclusion of L3 to a prediction equation with BW had a greater impact for females than it did for males,in terms of accounting for additionalvariance in elbow flexion strength.

The large contribution of limb length to the strength prediction equation for females may be explained by the relationship between the length of a muscle and the number of sarcomeres in series.21,22The number ofsarcomeres in parallel (physiological cross-sectional area)is proportional to the amount of tension that is produced whereas the number of sarcomeres in series(muscle fiber length)is proportional to the velocity at which tension is.23,24While the dependent measure in this study was mean torque,not velocity of shortening,it has been suggested that the number of sarcomeres in series,and therefore the length of a muscle,has a relationship with the amount of force being produced.22,25This relationship was demonstrated for sprint performance and leg characteristics in female sprinters.Abe and colleagues26found that increased fascicle length was highly correlated with increased shortening velocity and concurrently,sprintperformance.These physiological characteristics combined with females’decreased proportion of lean tissue mass may explain the large contribution of limb length compared to weight and circumference measurements.

4.2.Contribution of muscle activation

The contribution of muscle activation in addition to muscle size to the prediction ofstrength was assessed by incorporating RMS sEMG amplitude to equations consisting of BW and a second anthropometric variable.The addition ofsEMG RMS resulted in a significant(p＜0.05)increase in the varianceaccounted-for by each equation,except when the second variable was L3 for females.The minimal contribution may have been due to the immense contribution of L3 alone(partial R2=39.1%).Excluding this particular case,on average, sEMG RMS accounted for an additional10.1%of the variance in strength.Surprisingly,the addition of a third anthropometric variable instead of sEMG RMS resulted in superior prediction equations for both males and females.

The majority of the literature on force and sEMG is focused on the linear versus non-linear nature of the relationship,to create a calibrating equation throughout the range of muscle forces(0—100%maximal voluntary contraction).Musculoskeletalmodeling focuses mainly on individualmuscle forces, notmaximum force prediction atthe jointas accomplished in this study.27—29The impact of adding sEMG to a prediction equation for muscle force that already includes a measure of muscle size was less than expected.Hahn30used sEMG to predict isokinetic knee torque using a multiple linear regression.An equation containing limb position,height,body mass and sEMG produced R2values of 0.67—0.71.Similarly,Youn and Kim31used sEMG from the biceps brachii and brachioradialis for elbow flexion prediction and found correlations of 0.90 and above between observed and predicted forces.

One possible reason that sEMG had a greater contribution to the prediction of muscle strength in the aforementioned studies may be the inclusion ofactivity from multiple muscles, including antagonistic co-activation.Joint torque is the product of a multiple muscle system and we only included sEMG activity from the primary agonist.Praagman and colleagues32observed sEMG of elbow flexors and extensors during static contractions at varying joint angles and pronation—supination positions.They found that joint angle, moment arm,and muscle length influenced the EMG amplitude.Similarly,Brookham and colleagues33found that these same variables,and the load applied to the joint,influenced the amountof co-activation presentduring isometric contractions. The inclusion of sEMG from multiple muscles at different joint angles may be beneficial for the prediction of muscle strength.However,in agreement with the current findings, Hahn30reported that the primary force predictors for knee torque were the position of the limb,body mass and body height,followed secondarily by sEMG.

5.Conclusion

Anthropometrics provides a strong prediction equation for the estimation of isometric elbow flexion strength using multiple linear regression.While muscle activation,as measured by RMS sEMG activity,accounted for a significant (p＜0.05)amount of variance in most prediction equations, its contribution was comparable to the use of an additional anthropometric variable.Therefore,the hypothesis thatmuscle activation would improve the prediction equation more than anthropometrics alone cannot be entirely accepted.It was found thatthe strongestprediction equation forboth males and females included BW,forearm length,and elbow circumference.

Acknowledgments

This study was supported by the Natural Sciences and Engineering Research Council of Canada.This work is dedicated to the memory of Dr.Walter Kroll.

1.Gabriel D,Kamen G,Frost G.Neural adaptations to resistive exercise: mechanisms and recommendations for training practices.Sports Med 2006;36:133—49.

2.Staudenmann D,Poeleveld K,Stegeman D,van Dieen J.Methodological aspects of SEMG recordings for force estimation:a tutorial and review. J Electromyogr Kinesiol 2010;20:375—87.

3.Kroll W.Motor learning/control.In:Proceedings:NAPEHE,National Association for Physical Education in Higher Education:Theory into Practice:Annual Conference,Indianapolis,Indiana,January 6-9,1983. Champaign,IL:Human Kinetics Publishers;1983.

4.Ikai M,Fukunaga T.Calculation of muscle strength per unit crosssectionalarea of human muscle by means of ultrasonic measurement.Eur J Appl Physiol Occup Physiol 1968;26:26—32.

5.Sale D,MacDougall J,Always S,Sutton J.Voluntary strength and muscle characteristics in untrained men and women and male bodybuilders. J Appl Physiol 1987;62:1786—93.

6.Bamman M,Newcomer B,Larson-Meyer D,Weinsier R,Hunter G. Evaluation of the strength-size relationship in vivo using various muscle size indices.Med Sci Sports Exerc 2000;32:1307—13.

7.Mayhew J,Ball T,Ward T,Hart C,Arnold M.Relationships of structural dimensions to bench press strength in college males.J Sports Med Phys Fitness 1991;31:135—41.

8.Maganaris C,Baltzopoulos V,Ball D,Sargeant A.In vivo specific tension of human skeletal muscle.J Appl Physiol 2001;90:865—72.

9.Kroll W,Bultman L,Kilmer W,Boucher J.Anthropometric predictors of isometric arm strength in males and females.Clin Kinesiol1990;44:5—10.

10.Lippold OCJ.The relation between integrated action potentials in a human muscle and its isometric tension.J Physiol1952;117:492—9.

11.Bigland B,Lippold OCJ.The relation between force,velocity and integrated electrical activity in human muscles.J Physiol 1954;123:214—24.

12.Kleinbaum DG,Kupper LL,Nizam A,Muller KE.Applied regression analysis and other multivariable methods.4th ed.Boston:Duxbury Pr.; 2007.

13.DiCarlo S,Sipe E,Layshock P,Varyani S.Experiment demonstrating skeletal muscle biomechanics.Am J Physiol 1998;275:59—71.

14.Mayhew J,Piper F,Ware J.Anthropometric correlates with strength performance among resistance trained athletes.J Sports Med Phys Fitness 1993;33:159—65.

15.Hetzler R,Schroeder B,Wages J,Stickley C,Kimura I.Anthropometry increases 1 repetition maximum predictive ability of NFL-225 test for division 1A college football players.J Strength Cond Res 2010;24:1429—39.

16.Mayhew J,Jacques J,Ware J,Chapman P,Bemben M,Ward T,et al. Anthropometric dimensions do not enhance one repetition maximum predictions from the NFL-225 test in college football players.J Strength Cond Res 2004;18:572—8.

17.Bishop P,Cureton K,Collins M.Sex difference in muscular strength in equally-trained men and women.Ergonomics 1987;30:675—87.

18.Miller AEJ,MacDougall JD,Tarnopolsky MA,Sale DG.Gender differences in strength and muscle fiber characteristics.Eur J Appl Physiol 1992;66:254—62.

19.Kanehisa H,Ikegawa S,Fukunaga T.Comparison of muscle crosssectionalarea and strength between untrained women and men.Eur J Appl Physiol 1994;68:148—54.

20.Scanlan J,Ballmann K,Mayhew J,Lantz C.Anthropometric dimensions to predict 1-RM bench press in untrained females.J Sports Med Phys Fitness 1999;39:54—60.

21.Lindsey C,Makarov M,Shoemaker S,Birch J,Buschang P,Cherkashin A, et al.The effect of the amount of limb lengthening on skeletal muscle. Clin Orthop Relat Res 2002;402:278—87.

22.Rassier D,MacIntosh B,Herzog W.Length dependence of active force production in skeletal muscle.J Appl Physiol 1999;86:1445—57.

23.Narici M.Human skeletal muscle architecture studied in vivo by noninvasive imaging techniques:functional significance and applications. J Electromyogr Kinesiol 1999;9:97—103.

24.Wickiewicz T,Roy R,Powell P,Edgerton V.Muscle architecture of the human lower limb.Clin Orthop Relat Res 1983;179:275—83.

25.Burkholder T,Fingado B,Baron S,Lieber R.Relationship between muscle fibre type and sizes and muscle architectural properties in the mouse hindlimb.J Morphol 1994;221:177—90.

26.Abe T,Fukashiro S,Harada Y,Kawamoto K.Relationship between sprint performance and muscle fascicle length in female sprinters.J Physiol Anthropol 2001;20:141—7.

27.Challis J,Kerwin D.Determining individual muscle forces during maximal activity:model development,parameter determination and validation.Hum Mov Sci 1994;13:29—61.

28.Crowninshield R,Brand R.A physiologically based criterion of muscle force prediction in locomotion.J Biomech 1981;14:793—801.

29.Herzog W.Individualmuscle force estimations using a non-linear optimal design.J Neurosci Methods 1987;21:167—79.

30.Hahn M.Feasibility of estimating isokinetic knee torque using a neural network model.J Biomech 2007;40:1107—14.

31.Youn W,Kim J.Estimation of elbow flexion force during isometric muscle contraction from mechanomyography and electromyography.Med Biol Eng Comput 2010;48:1149—57.

32.Praagman M,Chadwick E,van der Helm F,Veeger H.The effectofelbow angle and external moment on load sharing of elbow muscles.J Electromyogr Kinesiol 2010;20:912—22.

33.Brookham R,Middlebrook E,Grewal T,Dickerson C.The utility of an empirically derived co-activation ratio formuscle force prediction through optimization.J Biomech 2011;44:1582—7.

Received 25 April 2012;revised 25 May 2012;accepted 28 May 2012

*Corresponding author.

E-mailaddress:dgabriel@brocku.ca(D.A.Gabriel)

Peer review under responsibility of Shanghai University of Sport

Production and hosting by Elsevier

2095-2546/$-see front matter Copyright?2012,Shanghai University of Sport.Production and hosting by Elsevier B.V.All rights reserved. http://dx.doi.org/10.1016/j.jshs.2012.05.004