An electromyographic analysis of hip abductor muscle activity when subjects are carrying loads in one or both hands.

Key Words: Electromyography, Hip abductors, Hip joint, Load carriage, Osteoartritis.

Many people with osteoarthritis of the hip strive to remain active and maintain a productive lifestyle. One activity required by a relatively active lifestyle is to walk while carrying objects such as a suitcase or a bag of groceries. Physical therapists often offer advice on various load carrying methods that, in theory, should minimize the magnitude of forces that cross the hip joint.(1,2) Minimizing these forces while carrying loads may decrease hip pain, thereby improving the likelihood that this functional activity could be continued.

In order to understand the relationship between carrying loads and the forces developed across the hip joint, the biomechanical role of the hip abductors during walking must be considered.(1-5) Frontal-plane rotary stability about the hip during single-limb support requires that the hip abductors generate a substantial torque to balance the torque produced by body weight (BW). The relatively large forces generated by the hip abductors are transferred across the hip, compressing the articular joint surfaces.(2,6-9) The total hip joint force produced during single-limb support may be as high as 3 to 3 1/2 times BW.(10,11) in Figure 1, the function of the hip abductors during single-limb support is shown. The force produced by the hip abductors is the single largest component of the total hip joint force produced during single-limb support. Reducing the need for excessive force production by the hip abductors should minimize the accompanying joint forces of muscular origin, which, in turn, should benefit the person with an unstable or painful hip.

Neumann and Cook(1) have used a noninvasive technique to estimate the relative force demands placed on the hip abductors and hip joints as nondisabled subjects walked while carrying external loads. Surface electromyography (EMG) was used to monitor the electrical activity from the hip abductors during the stance phase of walking as a load was carried by the side in one hand. Carrying a load contralateral to a given hip abductor muscle produced greater EMG activity than that produced during carrying the same load ipsilateral to that muscle. Neumann and Cook believed that greater EMG activity was indicative of higher hip abductor forces of muscular origin at the hip. interestingly, the ipsilaterally held load produced equivalent or less EMG activity than that produced while walking without a load. The authors explained these results on the basis of the relative position and length of the external moment arm of the loads (Fig. 2). The assumption was made that the stabilizing torque required for frontal-plane stability was divided between the hip abductor muscle and the ipsilaterally held load. The forces that cross the hip joint because of activation of the hip abductors would therefore be minimized. The authors believed their findings help explain why people with painful unilateral hip osteoarthritis often report that they can best manage a single load by carrying it on the same side as their painful hip.

Therapists often suggest that a single load be divided in half and carried bilaterally (ie, half of the load be carried by each hand).(2) The benefits of the bilateral method of carrying loads, however, have not been investigated. The purpose of this study was to verify the rationale of this clinical advice by comparing the hip abductor EMG activity produced by unilateral and bilateral methods of carrying loads in asymptomatic subjects. We felt that the data collected in this study would be generally applicable to patients with hip disease because the hip abductor muscles must still provide stabilizing forces during walking with hip pathology.

The null hypothesis of this study was that there would be no differences in the magnitude of normalized values for hip abductor EMG activity produced during unilateral and bilateral carrying of loads. Our hypothesis was limited to the events that occur during the "middle stance phase" of gait, defined for the purposes of this study as the period of single-limb support when the heel and toe of one foot are both in contact with the ground. Our hypothesis was limited to this period because this is the portion of the gait cycle on which our frontal-plane hip abductor model was based (Fig. 1).

The EMG measurements made from the hip abductors during load carrying were normalized to a percentage of the EMG voltage generated during walking without a load (%EMG). This normalized EMG value was used as a qualitative comparator of assumed relative force demands placed on the hip abductor muscle and hip joint. Two assumptions were made in the use of EMG data in this study. First, mechanical factors that affect a muscle's force output (eg, changing muscle length and moment arm distance) also alter the neural drive (reflected by EMG activity) to a given muscle.(12-14) The assumption was made in this study that hip abductor muscle length and moment arm distance during the middle stance phase of our normalization walks would be similar when subjects walked while carrying loads. There is no practical way to control muscle length and moment arm distance during a particular phase of walking. We assumed that any differences in EMG activity attributable to these mechanical factors would be minimal and insignificant. Second, Krebs et al(15) have performed an in vivo study showing that, in general, increasing the amount of hip abductor torque results in an increase in hip abductor EMG activity and in the pressure produced across the hip joint. The exact relationship between muscle force and surface EMG activity is not known for the hip abductors, and therefore no attempt was made in this study to use EMG values to predict absolute muscle or joint force magnitudes. By knowledge of the basic frontal-plane biomechanics of the hip abductors, the assumption was made that higher levels of EMG activity reflected relatively higher muscle forces and relatively higher associated muscularly generated hip joint forces. We believe this noninvasive EMG method of ranking different levels of presumed muscularly generated hip joint forces in people to be a valid and practical method.

In summary, surface EMG measurements were collected from the hip abductor muscles as nondisabled subjects walked on an indoor, hard-surfaced walkway while carrying different weight loads unilaterally and bilaterally. These EMG values were normalized to the level of EMG activity produced during walking without a load. Footswitches indicated to the computer the exact phase of each subject's gait.

Method

Subjects Thirty asymptomatic subjects (15 male, 15 female) were chosen for this study from a college student population. All subjects were paid for their participation in the study, and all subjects read and signed informed consent forms. To be selected for the study, subjects had to be considered healthy, without obvious postural asymmetry particularly related to a leg-length discrepancy of [equal to or greater than] 1.3 cm, scoliosis, or marked foot pronation deformity) or other musculoskeletal problems that required medical attention. Subjects ranged in age from 19 to 28 years (-)X=21.7, SD=2.0), in weight from 55.0 to 100.9 kg (-)X=66.5, SD=13.0), and in height from 1.6 to 1.9 m (-)X=1.7, SD=0.09).

Instrumentation

The EMG instrumentation used in this research has been described elsewhere.(14,16) Two surface on-site electrode assemblies were used with a ground electrode, a signal conditioning unit,(*) an oscilloscope, a microcomputer with an analog-to-digital convertor, and software for data collection and data reduction. Raw hipolar EMG data were processed by utilizing the root-mean-square (RMS) method, which produced a linear envelope or average EMG voltage over the time of a specified phase of gait.(16) The time constant used for processing the RMS EMG signal was 55 miniseconds. The raw EMG signal was monitored for movement artifact throughout all walking trials by use of an oscilloscope.

The RMS EMG signals from both right and left hip abductors were sampled at 100 times per second. Two EMG leads and one ground wire were attached to a small, lightweight

(*) Therapeutics Unlimited, 2835 Friendship St, Iowa City, IA 52240. plastic "distribution" box located on the midline of the subject's posterior lumbar region. A system of foot-switches worn by subjects produced unique voltages that allowed the computer to associate the EMG voltage level with a particular phase of gait.(1) Two on-off strip-type switches were attached to the bottom of each of a pair of flexible rubber galoshes. A rear-foot switch was taped to the heel of the undersurface of the footwear at a slight posterior-lateral to anterior-medial orientation about 30 degrees oblique to the sagittal plane.(1) A forefoot switch was similarly attached to the footwear's forefoot at an angle parallel with the rear-foot switch. During walking, both switches were either open (ie, during the swing phase) or closed (ie, during the stance phase). A combination of foot-switch closures defined the stance subphases of gait for each limb as follows: early stance phase-the interval from heel-strike to just prior to footflat, middle stance phase-the interval from footflat to just prior to heel-off, and late stance phase-the interval between heel-off and toe-off. Each footswitch was connected to a distribution box containing the electronics that determined the voltage output for a particular switch closure. The EMG and footswitch voltages traveled between subject and signal processor/computer via a single 12.2-m (40-ft) umbilical cable.

Pilot work performed in our laboratory showed that, on average, the middle stance phase of walking at a self-selected speed lasted about 0.4 second. Therefore, about 40 sample points of EMG activity per gait cycle were used to calculate the average EMG value during middle stance. This sample rate and subsequent number of data points were adequate to represent a general "moving average" of the hip abductor's activity.

Procedure

Preexperimental protocol. Each subject's height and weight were measured and shoe size ascertained. Following a brief description of the logistics of the study, subjects were led to a private room where EMG electrodes and a ground plate were applied. The skin over the posterior-lateral gluteal region was thoroughly cleansed with alcohol. To record the surface myoelectric activity of the hip abductors, electrodes were placed on the skin over the belly of the gluteus medius muscle bilaterally. The exact procedure used for application of the electrodes for hip abductor recording has been described in earlier studies.(1,14) Because of the high input impedance afforded by the EMG amplifier ([is greater than]25 M(Ohm)), measurement of skin impedance was not believed to be necessary. The ground electrode was placed on the skin over the anterior-medial aspect of the right tibia. Proper electrode placement was verified by palpation of the gluteus medius muscle and observation of a strong raw EMG signal from the hip abductors as the subject stood in single-limb support.

Subjects were led to the walkway, and the footswitch hardware was applied. Subjects practiced walking with various instrumentation in place until they reported they could walk "naturally" without feeling constrained by the wiring. As subjects walked, the experimenters sampled all data channels so that the EMG signal and the footswitch closure patterns could be reviewed on the computer screen. After verification of proper function of the various instrumentation, the gain to the EMG amplifiers were set so that the maximal EMG voltage was approximately 25% of the maximum voltage expected by the computer's analog-to-digital convertor.

The methods of carrying loads both unilaterally and bilaterally were demonstrated to the subjects. in both methods, a plastic container with a hinged handle on top, weighing 1.4 kg (3 lb) and having dimensions of 18X20x23 cm (7x8x9 in), was used. Unilateral load carrying required subjects to carry a single load with their right hand, much like one carries a single standard suitcase. The total loads for unilateral load carrying were 10% and 20% of each subject's BW. Bilateral load carrying required subjects to carry two loads, one at each side, much like one carries two suitcases. The total loads for bilateral load carrying were 10%, 20%, and 40% of each subject's BW (ie, 5%, 10%, and 20% of BW in each hand). These loads were selected because they covered the range of weights commonly transported by hand. The 40% of BW load was too heavy for most subjects to carry unilaterally, and the data therefore were excluded from the data analysis. During both unilateral and bilateral carries, normalized EMG measurements were obtained from both the right and left hip abductors.

Subjects were required to maintain a relatively constant self-selected walking speed during all walking trials. This allowed us to minimize any effects that changes in walking speed may have had on EMG activity. Subjects were instructed to walk at the same "slow, but comfortable, speed" that they chose while carrying the bilateral 40% of BW load. In our pilot work, we observed that, of all the load carrying conditions, the bilateral 40% of BW condition consistently resulted in the lowest self-selected walking speed. Therefore, by establishing a self-selected walking speed based on this condition, we believed that all other load carrying conditions could be performed at the required speed. After at least 2 minutes of walking, the average walking speed was calculated for three trials over a 10-m distance. The average self-selected walking speed over the 10-m walk was measured by stopwatch to the nearest 10th of a second. This walking speed, averaged over all 30 subjects, was 0.75 m/s. Subjects were taught to repeat all subsequent walks at a walking speed 10% greater or less than their own target walking speed. All subjects quickly learned to control their walking speed within our requirements.

Experimental protocol. The EMG voltages produced as subjects carried loads were normalized to a percentage of the EMG voltage produced during walking without a load (%EMG). To establish our preexperimental no-load EMG baseline voltage, footswitch and EMG data were sampled as subjects walked three trials at their self-selected walking speed. For each trial, data sampling began as subjects crossed a concealed 2-m mark on the walkway and continued for 10 subsequent seconds. Subjects were verbally instructed to stop walking after the completion of the data sampling. The EMG baseline voltage for each subject's phase of gait was determined by averaging the data across the three walking trials.

Data were next collected from each subject during five different load carrying conditions: Unilateral load carrying with loads of 10% and 20% of BW and bilateral load carrying with loads of 10%, 20%, and 40% of BW. The order in which subjects carried these loads was randomized. Subjects were allowed a 45-second rest period between each walking trial as experimenters displayed and verified the footswitch and EMG patterns on the computer screen. Data were accepted for analysis after the target walking speed was verified and a typical bilateral footswitch sequence was displayed on the computer screen. Subjects practiced one walking trial at each load carrying condition in order to familiarize themselves with the experimental procedure.

The data-collection procedure used for the load carrying experiments was performed exactly as described for the preexperimental no-load tests. The subjects' EMG data were sampled for 10 seconds for each of three walking trials for each of the five conditions. This experimental design provided data on approximately seven complete walking cycles per lower limb per walking trial. On average, each of the five load carrying conditions produced data from 21 complete gait cycles per limb.

Following the load carrying phase of the study, each subject established a postexperimental no-load EMG baseline voltage by repeating the preexperimental no-load walking trials. This procedure allowed us to establish intrasubject reliability of the EMG baseline measurements by comparing the EMG voltages produced before and after the load carrying experiments.

Data Analysis

Data collection provided us with normalized EMG measurements (ie, %EMG) for the right and left hip abductors for each load carrying condition for all 30 subjects. Each %EMG measurement represented the grand average of approximately 21 complete gait cycles per subject (ie, approximately seven gait cycles per trialxthree trials per load carrying condition). A multifactorial analysis of variance (ANOVA) on the normalized %EMG data from the unilateral load carrying trials during the middle stance phase of gait was performed first. The dependent variable for this preliminary analysis was %EMG from the reference hip abductor. As the unilateral load was always carried in the right hand, the right abductor %EMG data were analyzed for the ipsilateral load position and the left abductor %EMG data were analyzed for the contralateral load position.) The independent variables of interest were load (ie, 10% and 20% of BW) and load position (ie, ipsilateral and contralateral to the reference hip abductor). This analysis was followed by our main test of the null hypothesis. A multifactorial ANOVA was performed on the %EMG data collected during middle stance from both unilateral and bilateral load carrying conditions. The dependent variable was the hip abductor %EMG averaged over both right and left sides (%EMG[R+L]). The independent variables of interest were load (ie, 10%, 20%, and 40% of BW) and method of carry (ie, unilateral and bilateral carry). The dependent %EMG measurements for all experimental conditions were compared with each other and against 0% (ie, TABULAR DATA OMITTED the baseline EMG voltage produced during the preexperimental no-load condition) by using multiple a posteriori t-test comparisons with Bonferroni adjustments.(1,17) These adjustments maintained the a priori alpha level by dividing .05 by the number of comparisons made.

Results

To determine the relative intrasubject stability of the EMG baseline voltage throughout the experiment, we compared the grand mean EMG voltage (in millivolts) produced during the preexperimental no-load walking trials with that produced during the postexperimental no-load walking trials. Approximately 60 to 90 minutes separated these two measurements. Each preexperimental and postexperimental no-load grand mean EMG voltage was calculated by averaging all subjects' EMG measurements from both hip sides over all middle stance phases. The preexperimental no-load mean EMG voltage was 101.7 mV, and the postexperimental no-load mean EMG voltage was 93.3 mV. This 8% difference was considered small, and, from a practical standpoint, insignificant. This small change in the EMG baseline voltage did not have a systematic effect on the %EMG produced during load carrying, because the order of performance of all load carrying conditions was randomized for each subject. A Pearson Product-Moment Correlation Coefficient of .96 and an intraclass correlation coefficient (ICC[1,k]) of .94 were calculated for the preexperimental and postexperimental data (P[is not greater than].001).(17,18)

Table 1 shows the pertinent descriptive data for the %EMG produced by the hip abductors for both the unilateral and bilateral load carrying conditions. An ANOVA test on the mean %EMG data collected during the middle stance phase of unilateral load carrying revealed a statistically significant main effect for the variables of load and load position (Tab. 2). A statistically significant interaction was also demonstrated between these two variables. Figure 3 shows a plot of the hip abductor %EMG data for unilateral load carrying as loads were carried either ipsilateral or contralateral to the reference hip abductor. The post hoc test revealed that the values for %EMG produced by carrying the loads contralateral to the hip abductors were statistically greater than zero (ie, the EMG voltage generated during no-load walking). The values for %EMG produced by carrying the loads ipsilateral to the hip abductors were less than the no-load EMG baseline measurement. The four loadxload position group %EMG means shown in Figure 3 were all different from each other, except for the differences between ipsilateral 10% and 20% of BW conditions.

Table 3 shows the results of the ANOVA performed on the mean %EMG(R+L) data (ie, EMG values averaged across right and left hip abductors for both the unilateral and bilateral methods of load carrying). The data were analyzed during the middle stance phase only. Note that a statistically significant main effect and interaction was shown for and between the variables of load and method of carry. The plot in Figure 4 depicts the mean hip abductor TABULAR DATA OMITTED %EMG(R+L) for all method of carryxload experimental combinations. The post hoc test showed that the %EMG(R+L) produced by each of the five load carrying conditions shown in Figure 4 was greater than the no-load EMG baseline measurement. Furthermore, the mean %EMG(R+L) values from the five load carrying conditions were different from one another, except for the differences between the unilateral 10% of BW and bilateral 10% of BW conditions and between the unilateral 10% of BW and bilateral 20% of BW conditions. The alpha level for Bonferroni multiple comparisons(17) was set at .005 (ie, .05 divided by 10 loadxmethod of carry comparisons).

Discussion

Unilateral Load Carrying-Effect of Load Position on Hip Abductor Electromyographic Activity

The myoelectric activity of the hip abductors during the middle stance phase of walking was dramatically influenced by the position of the unilateral load carry (Fig. 3). These findings support our previous observation that people with a painful or unstable hip should, whenever practical, avoid carrying a load unilaterally contralateral to their pathological hip.(1) In sharp contrast, carrying a load ipsilateral to a pathological hip markedly minimizes the hip abductor EMG activity and presumably joint forces. The negative %EMG values shown in Figure 3 indicate that hip abductor EMG activity was less than that produced by walking without carrying a load. How the weight of an ipsilateral load carry can produce a percentage of the total torque required for frontal-plane stability about a particular stance hip is illustrated in Figure 2. Using the weight of the ipsilaterally placed load to partially substitute for and therefore minimize hip abductor contraction is a simple and practical method of "protecting" a given hip side. TABULAR DATA OMITTED

In Neumann and Cook's(1) original study of unilaterally carried loads, EMG data were collected and averaged over the entire stance phase of walking and then normalized to a percentage of the EMG activity generated during a maximal contraction. In our study, however, the EMG data were normalized to a percentage of EMG voltage produced during walking without a load and, perhaps more significantly, footswitches were used to indicate the hip abductor EMG level within each subphase of the stance phase. Figure 5 expanded the data set shown in Figure 3 by showing the hip abductor %EMG within each of the predefined subphases of stance. Although a detailed discussion of the hip abductor EMG activity in all subphases of stance was not an objective of this study, note the relatively strong negative effect the ipsilateral load had on the middle stance %EMG (Fig. 5). This finding is quite interesting because, for practically the entire duration of middle stance, the subjects' contralateral lower limb was involved in the swing phase. While walking without a load, the torque needed for frontal-plane stability about a hip joint involved in single-limb support would essentially be the sole function of the ipsilateral hip abductor (Fig. 1). The marked reduction in muscle activity during single-limb support while holding the load ipsilaterally can only be accounted for by the frontal-plane torque created by the weight and position of the load itself.

The ability to study the hip abductor's activity during the middle stance phase of unilateral load carrying revealed another subtle, but interesting, point. Specifically, the hip abductor %EMG during the middle stance phase was - 11.3% for the ipsilateral 10% of BW load and -17.3% for the ipsilateral 20% of BW condition. Although these differences were not statistically significant, the biomechanical model shown in Figure 2B would predict that the larger the ipsilaterally held load, the greater the reduction in hip abductor and myogenic hip joint force. Obviously, the validity of the concept has a limit. Eventually, an ipsilateral load of a certain weight would result in the need for ipsilateral hip abductor torque for frontal-plane stability.

Based on average estimated moment arm lengths (as depicted in Fig. 2B) and assumptions based on static rotary equilibrium,2,19 the limit was calculated to be at about 38% of BW. This calculation was a rough estimation of a relatively complex dynamic biomechanical event; however, it should serve as a point of reference for future analysis.

Unilateral Versus Bilateral Load Carrying-Effect of Method of Load Carry on Hip Abductor Electromyographic Activity

As stated previously, the data shown in Figure 4 were collected from bilateral and unilateral load carrying trials and were plotted as the %EMG(R+L). This bilateral abductor averaging would best reflect the "average" force of muscular origin imposed at either hip while carrying a load in either or both hands. As Figure 4 demonstrates, for loads of equivalent weights, the bilateral load carrying method appears to afford greater bilateral hip joint protection than does the unilateral load carrying method. The degree of reduction of the presumed hip force, however, was clearly dependent on the amount of the load. Of particular note was the fact that the bilateral 20% of BW load carrying condition produced less %EMG(R+L) (ie, 16.6%) than did the unilateral 20% of BW condition (ie, 40.6%).

People with hip disease should be aware that muscularly generated forces at their hip joints may be significantly reduced by using the bilateral load carrying method, especially with loads that weigh as high as 20% of BW. Doubling the weight of the load resulted in only a modest increase in bilateral hip abductor EMG activity, provided that the load carrying method changed from a unilateral to a bilateral strategy (Fig. 4). For example, the %EMG(R+L) produced by the bilateral 20% of BW load carrying condition was only a statistically insignificant 6% greater than that produced by the unilateral 10% of BW condition. As shown in Figure 4, the bilateral load carrying method is clearly the method of choice if one wishes to carry substantial weights while simultaneously minimizing the hip abductor EMG and associated joint forces.

We believe the differences in EMG activity (and presumed hip joint forces) between methods of carrying loads can be explained by the plots shown in Figure 6. The unilateral load carrying data (in either Figure 6A or 6B) demonstrate marked polarity in the algebraic sign, and large differences in magnitude of the %EMG, between right and left hip abductors. Because all unilateral loads were carried in the subjects' right hand, a very high positive left hip abductor %EMG and a slightly negative right hip abductor %EMG was expected.(1) The large "contralateral effect" of carrying a unilateral load skewed the %EMG(R+L) to a relatively high level. At both 10% and 20% of BW loads, the assumed muscularly generated forces (as reflected by %EMG[R+L]) from unilateral load carrying were much greater than the apparent forces secondary to bilateral load carrying. This contrast in hip abductor EMG activity between methods of carrying loads can be at least partially explained by observing the biomechanical model of bilateral 20% of BW load carrying (Fig. 7).

Our analysis is limited to the right hip abductor's force requirement during right single-limb support. Static frontal-plane rotary equilibrium requires a mathematical balance in clockwise and counterclockwise torques about the right hip. During the middle stance phase on the right hip, the contralateral load held by the left hand would combine with BW to create a relatively large clockwise torque about the right hip. Using the sample data shown in Figure 7, this combined clockwise torque was estimated to be about 720 in-lb. Without the 105-in-lb counterclockwise torque effect of the ipsilaterally carried load, the right hip abductors would, by necessity, be responsible for matching this entire clockwise torque. in theory, the addition of the 15-lb ipsilaterally carried load to this bilateral load carrying model reduces the internal (counterclockwise) torque demands placed on the right hip abductors from 720 in-lb to 615 in-lb. To estimate the actual reduction in joint force, the hip abductor torque is divided by an estimate of the length of the abductor moment ann (ie, 720 in-lb/1.73 in versus 615 in-lb/1.73 in). In theory, the addition of the 15-lb ipsilateral load reduces the right hip abductor force requirement from 416 to 355 lb, or by 61 lb. The theoretical net decrease in compressive force of 46 lb (ie, 61-lb muscle force reduction plus the 15-lb load increase) may seem insignificant. We believe, however, that it is a rational form of hip joint protection.

The model shown in Figure 7 can also be used to explain the high interside symmetry in hip abductor %EMG during bilateral load carrying demonstrated in Figure 6. The biomechanical events for the right hip abductor muscles during right stance phase have been described. The same points could have been made by emphasizing the left hip abductors during left stance. When considering an infinite number of gait cycles during bilateral load carrying, one can assume a near equivalency in inter-side abductor hip joint forces of muscular origin. This assumption is based on the similar %EMG values produced by the right and left hip abductors during bilateral load carrying Fig. 6). In this case, neither hip joint side would be subjected to a condition of exaggerated hip abductor joint forces because of muscle contraction.

Conclusions

We believe that people with hip disability should carry loads that are as light as possible and, if practical, divide the load in half and carry each load in a separate hand. Carrying loads bilaterally offers two specific advantages over unilateral load carrying for people attempting to protect their hips. First, when averaged over both hips, bilateral load carrying produced less hip abductor %EMG than did carrying an equivalent-weighing unilateral load, (This difference was statistically significant for loads that weighed 20% of BW.) A static frontal-plane biomechanical model allows the assumption that lower %EMG reflected a condition of lower muscularly generated joint forces. Lowering these forces may retard the clinical sequelae of a naturally degenerating hip as well as extend the life expectancy of a prosthetic hip replacement. Second, the interside symmetry of hip abductor %EMG during bilateral load carrying allows the assumption that the hip abductor forces would be nearly equivalent across hip sides. This is in sharp contrast to unilateral load carrying where apparent muscular forces differ considerably across hip sides and the side of high abductor muscle demand is a function of the side of the load. Handedness or job environment may bias a preferred side for carrying loads unilaterally, thus predisposing high hip joint forces over the hip located contralateral to the carried load. Bilateral load carrying, however, minimizes the likelihood of exposing either hip side to high muscularly generated forces.

Acknowledgments

We thank Ms Joan Holcomb from Marquette University's Instructional Media Center for her assistance with the illustrations. We also thank Mr Senthil K Natesan for his skill in computer programming and Mr Paul Rossman and Ms Amy Tubbs for their assistance with data collection.

References

1 Neumann DA, Cook TM. Effect of load and carry position on the electromyographic activity of the gluteus medius muscle during walking. Phys Ther. 1985;65:305-311.

2 Neumann DA. Biomechanical analysis of selected principles of hip joint protection. Arthritis Care and Research. 1989;2:146-155.

3 Inman VT. Functional aspects of the abductor muscles of the hip. J Bone joint Surg. 1947;29:607-619.

4 Johnston RC. Mechanical considerations of the hip joint. Arch Surg. 1973; 107:411-417.

5 Rydell N. Biomechanics of the hip-joint. Clin orthop. 1973;92:6-15.

6 McLeish RD, Chamley J. Abduction forces in the one-legged stance. J Biomech. 1970;3: 191-209.

7 Merchant AC. Hip abductor muscle force. J Bone Joint Surg [Am]. 1965;47:462-475.

8 Brand RA, Crowninshield RD, Johnston RC, Pedersen DR. Forces on the femoral head during activities of daily living. Iowa Orthopedic Journal 1982;2:43-50.

9 Maquet P. Biomechanics of the Hip: As Applied to Osteoarthritis and Related Conditions New York, NY: Springer-Verlag New York Inc; 1985.

10 Rydell N. Forces acting on the femoral head-prosthesis: a study on strain gauge supplied prostheses in living persons. Acta Orthop Scand 1966;37(suppl 88):1-132.

11 Brand RA, Crowninshield RD. The effect of cane on hip contact force. Clin Orthop. 1980; 147:181-184.

12 Neumann DA, Soderberg GL, Cook TM. Comparison of maximal isometric hip abductor muscle torques between hip sides. Phys Ther 1988;68:496-502.

13 Olson MA, Smidt GL, Johnston RC. The maximum torque generated by the eccentric, isometric, and concentric contractions of the hip abductor muscles. Phys Ther. 1972; 52:149-157.

14 Neumann DA, Soderberg GL, Cook TM. Electromyographic analysis of hip abductor musculature in healthy right-handed persons. Phys Ther. 1989;69:431-440.

15 Krebs DE, Elbaum L, Riley PO, et al. Exercise and gait effects on in vivo hip contact pressures. Phys Ther. 1991;71:301-309.

16 Soderberg GL, Cook TM. Electromyography in biomechanics. Phys Ther. 1984;64: 1813-1820.

17 SAS User's Guide: Statistics, 1982 Edition. Cary, NC: SAS Institute Inc; 1982.

18 Shrout PE, Fleiss J. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

19 Williams M, Lissner HR; LeVeau BF ed. Bio-mechanics of Human Motion. Philadelphia, Pa: WB Saunders Co; 1977.

DA Neumann, PhD, PT, is Assistant Professor, Program in Physical Therapy, Marquette University, Milwaukee, WI 53233-2269 (USA). Address correspondence to Dr Neumann.

TM Cook, PhD, PT, is Assistant Professor, Physical Therapy Education, The University of Iowa, 2600

Steindler Bldg, Iowa City, IA 52242. RL Sholty, PT, is Staff Physical Therapist, Physical Medicine and Rehabilitation, Columbia Hospital, 2025 E Newport Ave, Milwaukee, WI 53211. She was Research Assistant, Program in Physical Therapy, Marquette University, when this study was conducted.

DC Sobush, PT, is Associate Professor, Program in Physical Therapy, Marquette University.

This study was approved by the Human Subjects Review Committee at Marquette University and was funded through a research grant from the National Arthritis Foundation, 1314 Spring St NW, Atlanta, GA 30309.

This article was submitted January 15, 1991, and was accepted September 9, 1991,