Many studies have reported the effects of backpack mass on lower limb joint reaction force and torque, ground reaction forces (GRF), and braking and propulsive forces (4,5,12,14,16). Increases in GRF and lower limb joint reaction force associated with carrying heavy loads have been shown to increase the potential for injuries of the lower limb such as stress fractures, foot blisters, and localized muscle fatigue and soreness (3,9). In addition to lower limb injuries, carrying heavy loads may result in low-back pain and rucksack palsy. One study suggests back problems (pains and strains) account for up to 22% of all load-carriage injuries (10). Rucksack palsy is a traction injury to the brachial plexus characterized by numbness and lack of function in the fingers and hands. This injury is typically caused by carrying a heavy backpack supported solely by shoulder straps for an extended period of time (17).
One way of reducing the potential for rucksack palsy is to use a backpack that incorporates a frame and hip belt (1) to transfer a proportion of the forces generated by the backpack from the upper back and shoulder to the lower back (iliac crest and sacrum). However, the effectiveness of using a framed backpack for this purpose is unknown. The primary purpose of the present study was to determine the proportion of force that is transmitted through the hip belt when carrying a framed backpack and to determine whether backpack mass affects this proportion. The first hypothesis was that increasing backpack mass will result in an increase in the peak and average forces exerted by the backpack on the upper and lower back.
Kinoshita (7) showed that increasing backpack mass results in a proportional increase in GRF and concluded the increased GRF associated with carrying a load is due to the static effect of the load. Increasing the backpack mass increases the gravitational force exerted on the backpack center of mass (COM), and, consequently, the force the backpack is exerting on the lower and upper back during walking. Normalizing the forces exerted on the lower and upper back, and the backpack COM by the backpack’s weight provided a dimensionless value that was used to determine whether increasing backpack mass resulted in proportional changes in these forces. This allowed for a determination of whether or not the increase in force that was likely observed under larger backpack mass conditions was due to the static effect of the load. For instance, motion of the backpack in relation to the trunk would indicate the backpack and trunk acceleration time series are not completely in-phase, suggesting an exchange of force between these segments. Pierrynowski et al. (13) demonstrated heavier loads do not move in synchrony with the trunk during treadmill walking, suggesting there may be a change in backpack acceleration associated with increased backpack mass, which would, in turn, result in a disproportionate change in the total force acting on the backpack COM. An additional purpose of this study was to determine whether or not systematic manipulation of backpack mass results in proportional changes in the forces exerted on the backpack COM, and the lower and upper back. The second hypothesis was that increasing the mass of the backpack would result in forces exerted on the lower and upper back that are disproportionately greater than the increase in backpack mass.
Eleven healthy male subjects participated in the study (mean body mass: 85.87 kg ± 17.36 standard deviation, mean age: 22.45 yr ± 3.83, mean height: 1.79 m ± 0.11). Subjects were selected from a population of United States Army soldiers. None of the subjects had any orthopedic disorder or complicating medical history. Before participation, subjects gave informed consent in accordance with the policies of the United States Army Research Institute of Environmental Medicine (USARIEM) and the U.S. Army Medical Research and Material Command. The investigators have adhered to the policies for protection of human subjects as prescribed in Army Regulation 70-25, and the research was conducted in adherence with the provisions of 45 CFR.
The backpack used in this study was the same backpack used by United States Army called Modular Lightweight Load Carrying Equipment (MOLLE), which has an external frame made of a lightweight plastic polymer (Fig. 1, A and B). There were two points of contact between MOLLE and the carrier. The upper attachment point was comprised of shoulder straps that are attached to an upper-back pad, which makes contact solely with the anterior and superior surfaces of the shoulders and the upper back. The lower attachment point between MOLLE and the carrier was the hip belt, which makes contact with the iliac crest and sacrum (lower back). MOLLE’s hip belt was designed to only make contact with two points on the MOLLE frame and was offset from the rest of the MOLLE frame (Fig. 2). To test the effect of force exerted on the hip belt on the distance between the hip belt and frame, the backpack was attached to a manikin that represents the 50th percentile man. Anterior/posterior forces of up to 400 N were exerted on the backpack frame, and the distance between the frame and the hip belt was measured. The offset between the hip belt and the backpack frame was 4.3 cm when an anterior/posterior force of 400 N was applied to the frame. Under none of the force conditions did the hip belt make contact with the backpack frame other than at the attachment points.
To construct the force-sensing backpack (FSB), the MOLLE frame was fitted with two high-quality force transducers (AMTI, Watertown, MA) capable of measuring force in three dimensions. One force transducer was mounted at each of the two connection points between the hip belt and the backpack frame. These transducers were mounted in the frame such that the position of the hip belt was unaltered by their presence (Fig. 3). The force transducers were the connection point between the hip belt and backpack frame. This design allowed a measurement of the forces exchanged between the lower back and the backpack. To test the validity of the force readings from the FSB, known weights were suspended from the backpack while it was positioned in three orthogonal directions. The FSB demonstrated intraclass correlation coefficients of 0.997 (95% CI = 0.980–0.999) for validity (ICC 2,1) and 0.999 (95% CI = 0.9991–0.9997) for reliability (ICC 2,1).
The experiment consisted of three backpack mass conditions (13.6, 27.2, and 40.8 kg). The masses were selected to fall within the range of masses normally tested in backpack experiments (2,4,16). The backpacks were loaded such that the COM was in the same location for all conditions. Each backpack mass condition was comprised of a set of foam blocks cut specifically to hold the lead weight in the same location. The COM of the backpack was determined using an adapted balance board technique and a plumb line. The plumb line was aligned with the center of the force plate and raised vertically several feet. The backpack was put on the force plate (under the plumb line) and positioned such that the center of pressure read from the force plate was coincident with the center of the force plate. For a static object, this indicates the COM is also coincident with the center of the force plate. The plumb line (which had already been aligned with the center of the force plate) was then lowered until it “pointed” to the location of the COM of the backpack. The COM was marked with a magic marker, and 2-cm reflective marker was centered over the backpack COM. The rigid foam in the backpack prevented shifting of the load within the backpack bag and allowed for the assumption that the backpack moved as a rigid body.
One assumption of our model was that the backpack COM does not move in relation to the backpack frame. The backpack was packed with lead weight and rigid foam that completely filled the rucksack. Before the subject walked with the backpack, the location of the COM of the backpack was verified using the adapted balance board method described above. Backpack COM location was consistently in the same location, that is, the marker on the COM of the backpack was 2 cm in diameter, and the plumb line always pointed to the center of the marker. If the weight within the backpack shifted during data collection, the location of the COM of the backpack would have changed. The fact that the COM of the backpack did not change indicates the backpack did not violate the model’s assumption.
Data were collected in the Center for Military Biomechanics at USARIEM. Before data collection, anthropometric measures of hip and shoulder width were taken. Body mass and height were respectively measured using a SECA model 770 electronic scale (Germany) and a GPM anthropometer (Switzerland).
Position data were recorded via a seven-camera Qualisys Motion Capture system (Glastonbury, CT). Reflective markers (approximately 2 cm in diameter) backed with double-sided adhesive tape were placed on the right side of the volunteer’s body at the fifth metatarsal head, lateral malleolus, lateral femoral condyle, greater trochanter, ulnar styloid process, lateral epicondyle of the arm, acromion process, and temple. The subjects then donned the backpack. The hip belt and shoulder straps were adjusted to constant tension across subjects and backpack conditions using a tensiometer. In addition to the reflective marker that was placed on the COM of the backpack in the sagittal plane, two additional markers were placed on the backpack frame. These markers allowed a determination of the orientation of the FSB (and, therefore, the orientation of the force sensors) in the lab coordinate system (Fig. 4).
Subjects walked on a level treadmill at 1.34 m·s−1. Every subject walked with the backpack loaded for all three mass conditions. The sequence of backpack mass conditions was balanced across subjects. Subjects walked with each backpack for approximately 3 min. During the last 30 s, kinematic data were collected at 100 Hz, and kinetic data were collected (from the hip belt force transducers) at 1000 Hz. The computer collecting the kinetic data was used to trigger the computer collecting the kinematic data, thereby synchronizing the kinetic and kinematic data. This setup was tested before data collection; results indicated the data were synchronized.
The raw kinematic data were converted into 3-D data by means of the Qualisys system software. The use of a seven-camera system reduces the occurrence of missing data. If there were missing data during a stride, it would typically be on the order of three to five frames. Missing data were deduced using cubic spline interpolation, after which data from the total sequence were filtered at 6 Hz, using a second-order low-pass Butterworth software filter.
Sagittal plane thigh angle (the angle between the thigh and the vertical) was calculated from the interpolated and filtered time series. Peak thigh flexion was used to determine the beginning and ending of each stride. There were a total of 826 strides of data used in the final analysis (mean strides per subject and trial = 25.0 STD 1.3). All of the kinematic and kinetic data were separated into individual strides and time-normalized to percentage of stride.
The velocity and acceleration of the backpack’s COM were calculated from changes in the position of the backpack COM marker. The force on the backpack’s COM was calculated from the backpack’s acceleration, mass, and weight using Equations 1 and 2:
where FBPx and FBPz represent the anterior/posterior and vertical (respectively) force exerted on the backpack COM, aBPx and aBPz the anterior/posterior and vertical (respectively) acceleration of the backpack COM, mBP the mass of the backpack, and g the acceleration due to gravity.
The raw data from both of the force transducers in the FSB were low-pass filtered at 20 Hz. Hip force was calculated by multiplying the filtered data from each of the force transducers by the factory-provided calibration values. The total force exerted on the hips was calculated by summing the force exerted on the two transducers. To compare the forces from the backpack transducers to the force acting on the backpack COM, the forces from the transducers in the backpack were transformed from the local, “FSB” coordinate system to the global “lab” coordinate system. This was completed using a simple transformation described in Zatsiorsky (18).
The design of MOLLE allowed us to adopt a sagittal-plane model of load carriage that assumes two points of contact between the subject and backpack. The lower attachment point connected the lower back (iliac crest and sacrum) with the hip belt. The upper attachment between MOLLE and the carrier connected the upper back (including the superior and anterior surfaces of the shoulders) to the shoulder straps of the backpack. The model assumed any force exerted on the backpack COM that was not transmitted through the lower attachment point was transmitted through the upper attachment point. Consequently, the force exerted on the upper back and shoulders at every point in time was calculated as the difference between the forces exerted on the backpack COM and the forces exerted on the hip belt, using Equations 3 and 4:
where FUBx and FUBz represent the anterior/posterior and vertical (respectively) force exerted on the shoulders and upper back, and FLBx and FLBz represent the anterior/posterior and vertical (respectively) force exerted on the hip belt.
The proportion of vertical forces exerted on the hip belt and shoulders were calculated as the vertical force on the transducers divided by the total vertical force (Equations 5 and 6).
where FLB(%) and FUB(%) represent the proportion of the total vertical force exerted on the lower back and upper back, respectively, and FLBz, FLPz, and FUBz the vertical force exerted on the lower back, upper back, and backpack COM, respectively.
Dimensionless force variables were determined by dividing the actual peak vertical forces by backpack weight (Equations 7 and 8):EQUATION
where DimFLB, DimFUB, and DimFBP represent the dimensionless vertical force exerted on the lower back, upper back, and backpack COM, respectively, and PeakFLBz, PeakFUBz, and PeakFBPz represent the peak vertical force on the lower back, upper back, and backpack COM, respectively. A dimensionless value greater than 1 indicates the force is greater than the weight of the backpack, whereas a value less than 1 indicates the force is less than the weight of the backpack.
A one-way repeated measures analysis of variance (ANOVA) was used to test for a main effect of backpack mass (three levels, 13.6, 27.2, and 40.8 kg) on the dependent variables. If significant (P < 0.05) main effects were found, a Tukey honestly significant difference post hoc (P < 0.05), which should provide ample protection against a Type I error (15), was used to determine specifically which backpack conditions differed from each other.
As expected, increasing backpack mass was associated with an increase in the maximum and minimum vertical and anterior/posterior forces exerted on the lower back (sacrum and iliac crest), the upper back, and the backpack COM (Table 1). In addition, increasing backpack mass resulted in an increase in mean vertical force exerted on the lower back, upper back and shoulders, and backpack COM. The mean anterior/posterior force exerted on the lower back, upper back, and backpack COM was also significantly affected by backpack mass (Table 2).
Regardless of the backpack mass, the lower back supported approximately 30% of the weight, with the remaining 70% on the upper back (Table 3), that is, there was no statistically significant effect of backpack mass on the proportion of backpack weight supported by the lower back. In addition, there was no statistically significant effect of backpack mass on dimensionless lower-back force or dimensionless upper-back force. In contrast, there was a significant effect of backpack mass and dimensionless backpack COM forces. Increasing backpack mass resulted in an increase in vertical force exerted on the backpack COM proportionally greater than the increase in the weight of the backpack. Additionally, there was a significant main effect of backpack mass on the peak vertical acceleration of the backpack COM and on the range of vertical acceleration of the backpack COM (Table 4).
The first hypothesis was that increasing backpack mass would result in an increase in the peak and average forces exerted by the backpack on the upper and lower back; this hypothesis was supported. The second hypothesis was that increasing in the mass of the backpack would result in forces exerted on the lower and upper back that were disproportionately greater than the increase in backpack mass; this hypothesis was not supported.
This is the first study to quantify the proportion of force supported by the lower back during load carriage, and for the MOLLE pack, this proportion was not affected by backpack mass. One limitation of this study is that data were collected only on male subjects. Females typically have a broader sacrum resulting in a wider pelvis and more prominent iliac crest (11). The difference in the structure of the pelvis between males and females may influence the proportion of the backpack’s weight supported by the lower back, which is made up of the sacrum and iliac crest. Consequently, the current data set is limited in that it may not accurately illustrate the proportion of force carried on the lower back in female subjects. Work is currently underway replicating this study using female subjects.
As expected, increasing the mass of the backpack resulted in an increase in the magnitude of the peak and mean vertical and anterior/posterior forces exerted by the backpack on the lower and upper back. Dimensionless analysis revealed the vertical force the backpack exerted on the lower back and the vertical force exerted on the upper back increased proportionately to the weight of the pack. In contrast, the vertical force exerted on the backpack COM increased more than proportionately to the increase in weight of the backpack. The increase in dimensionless vertical force on the backpack COM observed in this study is due to a greater acceleration of the backpack COM when carrying a heavier pack. The peak vertical force exerted on the backpack COM occurs shortly after heel-strike and is likely the result of the backpack impacting the carrier’s upper and lower back. Similar to the trajectory of the body COM during walking (6), the trajectory of the backpack COM during walking is sinusoidal (Fig. 5). The backpack is raised and allowed to fall twice during each stride. The peak vertical force exerted on the backpack COM is likely related to the distance the backpack is raised and allowed to fall; increasing the distance the backpack is allowed to fall likely results in an increase in the peak vertical force. In the present study, the vertical amplitude of the backpack COM increased from 0.051 m when walking with the 13.6-kg pack to 0.054 m when walking with the 40.8-kg pack, possibly explaining why the increase in peak vertical force exerted on the backpack COM increases disproportionately to the mass of the backpack.
As expected, the mean vertical force per stride exerted on the backpack COM was similar to the weight of the backpack. That is, averaged across the stride, the net upward acceleration of the backpack COM was equal to the net downward acceleration of the backpack COM, and the vertical force exerted by the lower and upper back counteracted the weight + inertial force of the backpack. The calculation of the dimensionless forces gave insight into the peak forces experienced on the lower and upper back relative to the weight of the backpack. When the backpack reaches its lowest position, the vertical force exerted on the backpack to decelerate its downward movement and initiate upward movement reaches a dimensionless value of about 1.4 (DimFBP = 1.37–1.46), indicating the vertical force is slightly less than 140% of the weight of the backpack. In addition, the DimFLB was approximately 0.47, indicating the vertical force exerted on the lower back increased to about 47% of backpack weight, whereas DimFUB was approximately 0.97–1.02, indicating the vertical forces on the upper back and shoulders increased to about 95–102% of the backpack weight. Because the dimensionless variables are based on peak values (see Equations 7 and 8), it can be concluded the peak vertical force exerted on the lower back during load carriage is about 50% of the weight of the backpack, whereas the peak vertical force experienced by the upper back approximates the 100% the weight of the backpack. This was consistent across backpack mass conditions.
Because the mean anterior/posterior force exerted on the backpack, COM was less than 0.5 N; this force is likely not of operational relevance. In contrast, the mean anterior/posterior force exerted by the backpack on the lower back was anterior, and the mean anterior/posterior force exerted by the backpack on the upper back was posterior (Fig. 5). One reason for this may have to do with the location of the backpack COM. Because the COM of the backpack was posterior to the upper backpack attachment point during walking, the weight of the backpack generated a clockwise torque around this point (Reference Fig. 1a to orient angular motion). Additionally, due to the location of the backpack COM, vertical force generated by the lower back does not produce sufficient torque to counteract the torque generated by the weight of the backpack. Consequently, the upper attachment (upper back, and specifically, the anterior surface of the shoulders) needed to exert an anterior force that, in turn, generates a counterclockwise torque around the lower attachment point. This torque counteracted the torque caused by the weight of the backpack. However, because the mean anterior/posterior force exerted on the backpack COM was near zero, the lower back needed to exert a posterior force on the backpack to counteract the anterior force exerted by the upper back (and anterior surface of the shoulders) on the backpack. This interpretation suggests a trade-off in using a framed backpack. On one hand, the use of a framed backpack reduces the potential for rucksack palsy. On the other hand, the consistent anterior force exerted by the backpack on the lower back may contribute to lower back pain and soreness.
The notion that balancing torques results in the consistent anterior force exerted on the lower back and consistent posterior force on the upper back, suggests changing the location of the COM of the load may influence the anterior/posterior forces exerted on the upper and lower back. For instance, a double (back and front) pack shifts the COM of the load anteriorly compared with a backpack and, consequently, may result in a decrease in the peak and mean anterior/posterior forces exerted by the pack on the lower back. This, in turn, may decrease the occurrence of low-back pain associated with load carriage. Previous research comparing ratings of pain, soreness, and discomfort between a backpack and a double pack substantiate this claim. The use of a double pack results in a decrease in low-back pain while carrying heavy loads (8).
Data on the peak and mean forces exerted on the backpack COM, lower back, and upper back were presented in this manuscript. The peak force data were presented to provide information on the maximum and minimum forces exchanged between the carrier and the backpack, and consequently, provide information on the range of forces experienced by the carrier. Peak force values only provide information on a very small part of the stride cycle. In contrast, the mean data is presented to provide information on the typical amount of force exchanged over the entire stride cycle. Reporting both values (peak and mean) was important in characterizing the effect of backpack mass on the distribution of forces between the carrier and the backpack in that the peak and mean force provide different information.
The current study used a novel approach to measuring the distribution of forces between the upper and lower back during load carriage. The results clearly demonstrated that the use of an external frame backpack with a hip belt transfers approximately 30% of the vertical force generated by the backpack to the lower back (iliac crest and sacrum). Although it appears certain that transferring some of the load from the upper back to the lower back is advantageous, information is lacking as to what percentage of load distribution between the upper and lower back is optimal in terms of reducing the potential for injury resulting from high forces, and how this proportion may influence metabolic cost. Future research should focus on how variation in the proportion of weight supported by the upper and lower back during load carriage affects comfort, injury risk, and metabolic cost, so as to determine an optimal load distribution between upper and lower back.
1. Bessen, R. J., V. W. Belcher, and R. J. Franklin. Rucksack paralysis with and without rucksack frame. Mil. Med. 152: 372–375, 1987.
2. Charteris, J. Comparison of the effects of backpack
loading and of walking speed on foot-floor contact patterns. Ergonomics 41: 1792–1809, 1998.
3. Giladi, M., C. Milgrom, Y. Danon, and Z. Aharonson. The correlation between cumulative march training and stress fractures in soldiers. Mil. Med. 150: 600–601, 1985.
4. Harman, E. A., K. H. Han, P. Frykman, and C. Pandorf. The effects of backpack
weight on the biomechanics of load carriage. U.S. Army Research Institute of Environmental Medicine Military Performance Division Technical Report T-00-17, Natick, MA, 2000.
5. Harman, E. A., K. H. Han, P. N. Frykman, M. Johnson, F. Russel, and M. Rosenstein. The effects of gait timing, kinetic and muscle activity of various loads carried on the back. Med. Sci. Sports Exerc. 24: S129, 1992.
6. Inman, V. T., H. J. Ralston, and F. Todd. Human Walking. Baltimore: Williams & Wilkins, 1981, pp. 168.
7. Kinoshita, H. Effects of different loads and carrying systems on selected biomechanical parameters describing walking gait. Ergonomics 28: 1347–1362, 1985.
8. Knapik, J., P. Ang, H. Meiselman, et al. Solider performance and strenuous road marching: influence of load mass and load distribution. Mil. Med. 162: 62–67, 1997.
9. Knapik, J., E. A. Harman, and K. Reynolds. Load carriage using packs: a review of physiological, biomechanical and medical aspects. Appl. Ergon. 27: 207–216, 1996.
10. Knapik, J., K. Reynolds, J. Staab, J. A. Vogel, and B. Jones. Injuries associated with strenuous road marching. Mil. Med. 157: 64–67, 1992.
11. Moore, K. L. Clinically Oriented Anatomy, 4th Ed. Baltimore: Williams & Wilkins, 1992. pp. 1164.
12. Obusek, J. P., E. A. Harman, P. N. Frykman C. J. Palmer, and R. K. Bills. The relationship of backpack
center of mass location to the metabolic cost of load carriage. Med. Sci. Sports Exerc. 29: S205, 1997.
13. Pierrynowski, M. R., R. W. Norman, and D. A. Winter. Mechanical energy analyses of the human during load carriage on a treadmill. Ergonomics 24: 1–14, 1981.
14. Polcyn, A. F., C. K. Bensel, E. A. Harman, J. P. Obusek, C. Pandorf, and P. Frykman. Effects of weight carried by soldiers: combined analysis of four studies on maximal performance, physiology and biomechanics. U. S. Army Soldier and Biological Chemical Command, and U. S. Army Research Institute of Environmental Medicine Technical Report, Natick, MA, TR-02/010, 2000.
15. Portney, L. G., and M. P. Watkins. Multiple comparison tests. In: Foundations of Clinical Research, Applications to Practice, M. C. Davis (Ed.). East Norwalk, CT: Appleton and Lange, 1993, pp 397–418.
16. Vacheron, J. J., G. Pomarat, R. Chandezon, and G. Vanneuville. The effect of loads carried on the shoulders. Mil. Med. 164: 597–599, 1999.
17. Wilson, W. J. Brachial plexus palsy in basic trainees. Mil. Med. 152: 519–522, 1987.
18. Zatsiorsky, V. M. Kinematics of Human Motion. Champaign, IL: Human Kinetics, 1998, pp, 419.