The deadlift exercise is a commonly prescribed exercise for athletes in all sports because of its involvement of most of the lower body musculature (1) and its importance as a basic component of several other commonly prescribed power lifts such as the power clean. Like most full-body exercises, however, significant time must be spent by the athlete learning proper technique to avoid musculoskeletal injuries. The low back in particular is a common injury site, even in well-trained individuals, due to the high forces generated in the lumbar area. A 6-year study of elite powerlifters by Calhoon et al. (3) indicated that 23% of all injuries during that time were low back injuries, 74% of those being muscle strains. Another study by Keogh et al. (14) reported that in 101 lifters of different ages and genders, 23.7% of all injuries were low back injuries, with the deadlift exercise alone accounting for 11.9% of all injuries. These rates were similar to what was found in a retrospective study of strongman athletes for whom the low back accounted for 24% of all injuries, with the majority (40%) of those being muscular in nature (17).
It is likely that injury rates are even higher in inexperienced lifters with less background in proper technique. Data from Escamilla et al. (8) reported that less-skilled lifters have a wider stance, forcing a wider grip and a more stooped posture in the initial stage of the lift. This created a greater stress on the low back (longer lever arm), which can increase injury risk in the lower skilled participants.
Alternate exercises that place less stress on the lower back while athletes are learning proper technique and developing necessary muscle strength would be highly desirable to minimize injury risk. For example, in contrast to the conventional deadlift, the sumo style deadlift uses a significantly wider “toe-out” stance which allows the arms to be placed inside the knees. The result is a narrower grip, with the trunk significantly more upright before the lift (7,8,15), and the bar closer to the body during the lift (5), and a resulting reduction of joint moments and shear forces (5). Despite the advantages of the sumo lift cited above, anecdotal evidence suggests that conventional lift is taught more frequently because it is part of the learning progression for power lifts.
A so-called “walk-in” deadlift machine uses independent handles rotating on a fulcrum several feet in front of the lifter, allowing the subject to position the load on either side of the body instead of in front of the legs (Figure 1). Theoretically, this would allow better alignment of the load with the body's center of gravity throughout the range of motion, allowing a more upright torso position and decreasing the lever arm for the load. Measurement of body positioning, along with electrical activity of lower back muscles would help to answer the question of whether this machine would be a safe training alternative for the deadlift for inexperienced, rehabilitating, or delinquent lifters, but to date no studies have measured muscle activity or joint angles with the “walk-in” deadlift machine. The purpose of this study is to compare body posture and electromyographical (EMG) activity in more experienced and less experienced weight lifters during the conventional deadlift and several positional variations of the “walk-in” deadlift machine exercise. We hypothesized that the use of the “walk-in” deadlift will reduce the involvement of the erector spinae muscles due to a more upright posture.
Experimental Approach to the Problem
The primary purpose of this study was to determine differences in low back muscle activity and body posture between the “walk-in” deadlift (WID) and the conventional deadlift (CDL). To that end, EMG and motion capture analysis was carried out on subjects performing both lifts in random order. Because the WID uses a fixed lever with the load moving through an arc, minor changes in foot and hand positions have the potential to affect the relationship between the body's center of gravity and the load. Therefore, 2 different foot positions for the WID were used in the comparison. Electromyographical analysis of lower body muscles was carried out to determine whether WID and CDL in fact involved the same muscles and could be used interchangeably, and video analysis was used to discover whether the WID resulted in a more desirable upright posture. The secondary purpose was to determine if highly-skilled deadlifters would assume more optimal body position during the lift than lower-skilled deadlifters, and the group comparisons were made to answer this question. Because many conditioning programs involve the deadlift exercise, off-season male and female athletes from a variety of sports, were recruited. Lower-skilled deadlifters were recreationally active weight lifters with little experience in the deadlift exercise who were not currently using the exercise.
The study protocol was approved by the University of South Carolina Institutional Review Board. Before enrolling in the study, subjects were informed of the risks and benefits of involvement and signed an approved informed consent form. Fifteen active weightlifters (2 females, 13 males) between 18 and 24 years old with no recent injury history were recruited through word of mouth, flyers, and email solicitation. All were students at the University of South Carolina Upstate. Subjects were divided into high-skilled (HS) and low-skilled (LS) groups for purposes of the analysis. Mean height and weight for the HS group were 1.78 m ± 0.097 (SD) and 78.9 kg ± 4.4, respectively, which were not significantly different from those of the LS group 1.79 m ± 0.035 m and 84.2 kg ± 12.3, respectively. Both females were in the HS group.
After signing an informed consent form, subjects reported to the Spartan Performance Center on the USC Upstate campus on 3 separate occasions. On the first occasion, subjects were tested to find their conventional deadlift 3-repetition maximum (3RM) using well-known procedures (11) with a 20 kg Diamond Pro Olympic barbell (Hartselle, AL, USA) with a 32 mm shaft diameter and rubber bumper plates (Troy Barbell, Houston, TX, USA). All testing was conducted by a Certified Strength and Conditioning Specialist. The 3RM procedure consisted of a warm up with a light weight followed by 3–4 sets with progressively increasing weight culminating with an attempt at the expected 3-repetition maximum. On the second occasion, no sooner than 48 hours after the first, subjects were tested to find their 3RM on the Nautilus XPLoad Deadlift and Shrug machine (Vancouver, WA, USA) with the machine handles set to the lowest height to mimic the height of an Olympic bar with 45 cm diameter disc weights. After determination of 3RM, subjects were separated into skill groups based on their experience with the deadlift exercise: subjects who had been taught advanced deadlift technique by an experienced strength coach, and who had been consistently using the deadlift in their current workouts for more than 6 months were classified as HS (n = 8), whereas low-skilled (LS, n = 7) subjects were not using the deadlift and had not received deadlift instruction from a trained instructor. To avoid potential risk from the exercise, all subjects were required to have been carrying out regular (at least 3 days per week) recreational or supervised strength training.
On the third and final visit to the lab, electrical activity of 4 muscles was monitored using EMG during submaximal deadlifts with both the weighted bar and the walk-in deadlift machine. After the preparation for EMG data collection, subjects performed a maximal contraction against immovable resistance at a specified joint angle for each muscle. The data collected during each attempt were recorded as maximal electrical activity of the involved muscles and were used to normalize results from the upcoming submaximal lifts. For the erector spinae (ES), the subjects were placed in a prone position and attempted to raise their torso with their hands at their side. For the gluteus maximus (GM), standing subjects leaned over a waist high bench with their leg straight and attempted to extend their hip. For the biceps femoris (BF) muscle, subjects maintained the same bent over posture used in the hip extension, but attempted to flex their knee from a position of 90° of flexion. Lastly, for the vastus lateralis (VL), subjects were seated on the bench and attempted to extend the knee from a position of 90° of flexion. After a brief rest, subjects completed, in random order, 3 repetitions each of 3 different exercises at 80% of the 3RM for that exercise, separated by a 3-minute rest period. This load was chosen to imitate normal training loads. Participants were given only basic safety instructions, but to avoid confounding the examination of differences between HS and LS participants, no specific technique instruction was provided. The exercises were (a) conventional straight-bar deadlift (CDL), (b) WID with the machine's handles aligned with the end of the toes and held with a pronated grip (ToePro), and (c) WID aligned with the ball of the foot and held with a pronated grip (BallPro) (Figure 1). At the conclusion of the final set, subjects were released and allowed to remove electrodes at their own discretion. Electromyographic activity, expressed as % maximum activity, was later analyzed and compared between the CDL and the 2 variations of the WID.
High-Speed Video Data Collection and Analysis
During the third visit to the laboratory, subjects had reflective markers placed on their shoulder (acromion process), hip (greater trochanter), knee (lateral side, between distal femur and proximal tibia), and ankle (lateral malleolus) joints. A high-speed video camera (Basler, Ahrensburg, Germany) captured images at a rate of 120 frames per second, which was then analyzed by motion capture software (Innovision Systems, Columbiaville, MI, USA) after calibration with a calibration frame of known dimensions. Trunk and knee angles were reported at 3 separate lift phases: (a) start—the frame when the hands began to move vertically, (b) midconcentric—the frame closest to when the hands were horizontally on level with the knee marker as the bar or machine lever was travelling upward, and (c) mideccentric—at the frame closest to when the hands were horizontally on level with the knee marker as the bar or machine lever was being returned to the starting position. Trunk angle was reported in absolute terms as the number of degrees clockwise from the left horizontal axis, with a greater value indicating a more upright torso. Knee angle was reported as the relative angle between the shank and the thigh, with a lower value indicating a more flexed knee. All subjects were facing to the left in the video analysis, and to maximize marker visualization throughout the lift, the single camera was placed at a position 10° posterior to the line between the back of the subjects' heels. Pilot data showed that this slightly off-center camera position did not significantly affect the accuracy of joint angles, and the camera position remained the same for all subjects.
Electromyographical Data Collection and Analysis
Electromyographic data were recorded using a Biopac data collection system (Goleta, CA, USA). Electrodes were placed while the subject was in a standing position after recommendations of the SENIAM project (13). A reference electrode for each muscle was placed on a bony prominence nearby. Before testing, the skin surface over the center of the muscle belly was shaved and cleaned vigorously with an alcohol pad to optimize the strength of the EMG signal. Once preparation was complete, signal integrity was checked by having subjects contract against resistance with a movement involving each tested muscle.
Electromyographical signals were filtered using a bandwith of 20–450 Hz with the gain set to ×1,000. Samples were collected at 1,024 Hz and stored for later analysis. Root mean square values for each muscle were then normalized by dividing EMG activity from CDL and WID exercises by the processed signals collected during the maximal isometric efforts. A recent review of EMG normalization methods reports that the use of EMG activity during maximal isometric contractions at an arbitrary joint angle produce interindividual reliability that is at least as high as with other methods (2). Intraclass correlation coefficients for surface EMG measurements during maximal isometric, submaximal concentric, and submaximal eccentric contractions have been shown to have intratester reliability of 0.93, 0.87, and 0.96, respectively (9).
A 2-way Repeated Measures analysis of variance was used to compare deadlift type as the within-subjects factor and skill level as the between-subjects factor. When differences were identified, 2-way paired t-tests were used post hoc, with significance level set to p ≤ 0.05 for all comparisons. For determination of differences in trunk segment lengths at the beginning and end of the concentric phase, a 2-way pairwise t-test was used, with the alpha value also set to 0.05. Effect sizes (d) were calculated and indexed as small (d < 0.5), medium (d ≥ 0.5, but <0.8), and large (d ≥ 0.8) (6). For body weight, 3RM, and 3RM/BW, independent 2-tailed t-tests were used to compare the means. IBM SPSS version 22 was used for all statistical analyses.
The mean CDL 3RM was 129.2 kg ± 23.6 for combined skill groups (127.9 kg ± 13.9 for LS subjects and 130.5 kg ± 34.3 for HS subjects). The mean WID 3RM was 195.3 kg ± 46.8 (183.7 ± 38.3 for LS subjects and 206.8 ± 60.4 for HS subjects). There were no significant differences between skill groups for the same lift, but the WID 3RM was significantly higher than the CDL 3RM (p ≤ 0.001) in the combined groups. The mean CDL 3RM per kg body weight for combined skill groups was 3.46 ± 0.64 (3.37 ± 0.62 for LS subjects and 3.54 ± 0.71 for HS subjects), whereas the mean WID 3RM per kg body weight for the combined groups was 5.15 ± 1.15 (5.01 ± 1.15 for LS subjects and 5.29 ± 1.22 for HS subjects). There were no significant differences between skill groups for the same lift, but the WID 3RM per kg BW was significantly higher than the CDL 3RM per kg BW (p ≤ 0.001).
There were no differences between skill groups for either body position during the lifts or for EMG activity. In the combined skill groups, video data showed trunk angle (Figure 2A) at the start of the lift to be 23.7° ± 11.3 SD for the CDL. That angle was significantly greater (p ≤ 0.05) for both the BallPro (29.9° ± 12.0, d = 0.47) and the ToePro (32.4° ± 10.4, d = 0.75). In the midconcentric phase, trunk angles were 42.7° ± 3.7 for the CDL, 46.0° ± 10.1 for the BallPro, and 46.9° ± 6.8 for the ToePro, with only the ToePro shown to be significantly different from CDL (p ≤ 0.05, d = 0.65). For the mid-eccentric phase of the lift, CDL (42.9° ± 6.5) and BallPro (42.0° ± 9.1) trunk angle were the same, but ToePro was significantly greater, at 47.2° ± 7.0 with an effect size of 0.61.
For the knee (Figure 2B), CDL (110.8° ± 11.5) and BallPro (104.7° ± 13.1) were similar at the starting position, but ToePro knee angle was significantly less (101.6° ± 10.6, d = 0.93). In the midconcentric phase, both BallPro (135.7° ± 14.2, d = 2.70) and ToePro (136.5° ± 8.8, d = 1.97) were significantly less than CDL (159.3° ± 5.9). Similarly, in the mideccentric phase, both BallPro (129.2° ± 14.0, d = 2.48) and ToePro (127.7° ± 8.9, d = 1.68) were significantly less than CDL (150.5° ± 7.7).
For the combined groups, minimum trunk length at the start of the concentric phase of the lift (Table 1) was 0.420 m ± 0.028 for the CDL, which was significantly shorter (p ≤ 0.05) than the same measurement for ToePro (0.474 m ± 0.036, d = 1.59) and BallPro (0.473 ± 0.034, d = 1.66). Trunk length for all 3 exercises was significantly higher at the end of the concentric phase of the lift compared with the beginning (CDL 0.53 ± 0.033, ToePro 0.60 ± 0.058, and BallPro 0.593 ± 0.047), with ToePro and BallPro trunk lengths significantly higher than CDL in this position (p ≤ 0.05, d = 1.442 and 1.506, respectively). There were no differences across skill groups.
In the combined low- and high-skilled groups, electrical activity of the ES during the BallPro variation (53.1% ± 8.6, effect size 0.69) was significantly lower than that during CDL (73.19% ± 6.9), but was statistically unchanged for ToePro (58.0% ± 7.5) (Figure 3). Electrical activity of VL was significantly higher in both BallPro (79.9% ± 7.7, d = 1.41) and ToePro (64.3 ± 5.7, d = 0.89) compared with that in CDL (48.6% ± 3.4), whereas GM activity was lower for BallPro (30.1% ± 3.0, d = 0.98) and ToePro (30.2% ± 4.4, d = 0.87) than for CDL (47.1% ± 5.8). Biceps femoris activity was unchanged for all 3 exercises. There were no differences in muscle activity of low-skilled subjects compared with high-skilled subjects in any of the 3 lifts.
Although the skill level had no effect on either body position or EMG activity, a number of differences were found between the CDL and the WID variations in the combined groups, most of which exhibited large effect sizes, defined as d = 0.8 or greater (6). In the BallPro lift vs. CDL, for example (Figure 2A), trunk angle at the start of the lift was greater (torso more upright) and the knee angle was less than (more flexed) in the midconcentric phase. These differences were also present during the eccentric phase. For the ToePro lift, the subjects exhibited a more upright posture than for CDL at the start, midconcentric, and mideccentric phases of the lift. The knee was also more flexed in all phases compared with CDL.
Swinton et al. (16) compared the CDL with a hex bar deadlift, which, like the WID, also shifts the load posteriorly compared with CDL. As in the present study, they found that knee flexion in the starting position was significantly greater during the hex bar lift, but unlike our study, did not find differences in the angle of the torso. This may have been due to the skill level of their subjects compared with those in the current study. Our subjects were college level athletes and recreational lifters, some of whom had practiced and been instructed in the deadlift, but had not focused on technique or maximal lifting, whereas Swinton et al. recruited experienced powerlifters, whose training likely allowed them to maintain an upright torso even when executing the CDL. A comparison of knee and back angles for the starting position of the CDL illuminates the effects of advanced training, with the subjects from Swinton et al. (16) achieving a trunk angle of 55.2° in comparison with 23.7° in our combined groups, and knee angle of 72.5° in comparison with our result of 110.8°.
Camara et al. (4) compared CDL and hex bar EMG activity at 65 and 85% 1RM, and found that VL activity was higher with the hex bar lifts, whereas ES and BF activities were higher for the CDL (when the 65 and 85% lifts were combined). These results are similar to our findings comparing WID and CDL, with the exception that BF activity was not affected by lift type in our study. Although the hex bar and WID both allow for more upright posture (16), there seem to be differences in the lifts as well. For example, Camara et al. found no difference in the 1RM between the CDL and the hex bar for their subjects, whereas our subjects had a significantly higher mean 3RM on the WID.
To gain an understanding of how posture affected stress on the low back, hand position in relationship to the knee at the moment of the bar passing the knee was also measured. Although this was not a direct measurement of the lever arm for the load in relation to the hip joint (because there was no measurement of the center of gravity), a posterior shift in hand position from its position during the CDL would be expected to decrease the resistance-generated hip joint moment. We found that in comparison to CDL, the hands were shifted significantly posteriorly (p ≤ 0.05) for both BallPro and ToePro at midconcentric lift, and more so for the BallPro compared with ToePro. Viewed in light of previous findings with the hex bar deadlift, this likely indicates higher joint moments at the hip and spine when performing the CDL. Swinton et al. (16) compared peak joint moments during the CDL and a hexagonal bar deadlift and found spine moments to be higher in the CDL at intensities ranging from 10 to 60% 1RM and hip moments to be higher in the CDL up to 70% 1RM. Knee peak moments, however, were higher in the hex bar lift at intensities up to 80% 1RM (16).
In support of the hypothesis, postural differences were accompanied by altered EMG activity (Figure 3), but not always as expected. Electromyographical activity was significantly lower in the erector spinae during the BallPro exercise compared with the CDL (albeit with a medium effect size of 0.69), but was not different during the ToePro exercise, even though both machine variations showed a greater trunk angle than CDL. This disparity may have been due to the longer estimated lever arm for the load during ToePro, based on the in hand position differences discussed earlier and the shift in peak joint moments with deadlift alternatives described by previous research (16). If there was a longer lever arm, it would explain the higher ES muscle activity despite the more upright posture in ToePro compared with BallPro.
During both BallPro and ToePro, muscle activity was significantly lower in the gluteus maximus and higher in the VL with large effect sizes, underlining a clear difference in the muscle activation pattern in the WID compared with the CDL (Figure 3). The most direct comparison of these results with previous research might be studies examining the front squat vs. the back squat, where it is often assumed that the knee extensors are more active during front squat due to the more upright posture of the lifter and the purported shift of the load in relation to the center of gravity (10,18). However, studies have exhibited some disagreement as to whether there are differences in muscle activity between the 2 exercises; one showed no differences using a 70% 1RM load (10) and another found higher activity in the vastus lateralis in the front squat (small effect size), and higher semitendinosus activity during the ascending phase of the front squat (moderate effect size) (18). Both studies, however, showed kinetic and kinematic differences, including higher compressive forces on the knee with the back squat (10), and a greater trunk lean during the back squat (18). The latter result provides some support for the current study's finding of differences in trunk angle with seemingly minor adjustments to an exercise. Caution should be exercised, however, when comparing results from squat studies with those from the deadlift, because the squat and deadlift have been shown to have different kinematic profiles (12).
Trunk length was significantly longer (p ≤ 0.05) in both WID lifts compared with CDL at the start of the lift and at the end of the concentric phase, indicating better maintenance of lumbar lordosis through the lift (Table 1). Hales et al. (12) made the same comparison between the deadlift and the squat in trained powerlifters, and found that it was more difficult to maintain proper lumbar lordosis during the deadlift than the squat as indicated by shorter trunk lengths. Although the purpose of that study was different from ours, it serves as confirmation of the postural difficulties of the standard deadlift, even in experienced lifters.
The fact that there were no differences between the HS and LS groups was interesting, although not completely surprising. Out of a desire to minimize potential risk to subjects, only those with a consistent history of resistance training were chosen to participate, and perhaps that general training experience was transferred as an ability to select an appropriate lifting technique even without explicit instruction. Alternately, it is possible that subjects in the HS group had not engaged in enough technique training to make a significant skill difference, because most athletes are using the CDL as a small part of their overall training regimen rather than as a specific performance goal (as with powerlifters, for example). Additionally, the HS tested were athletes from a moderately sized NCAA Division I program and participated in a variety of sports, including, track, softball, and basketball, or were simply well-trained recreational lifters using the deadlift consistently. Using a more homogenous group of subjects, such as powerlifters or athletes from the same sport might have resulted in a greater difference in skill level between the groups.
The study had several limitations. Firstly, the criteria to distinguish low-skilled vs. high skilled subjects could have been more disparate, perhaps setting a longer time limit on the date of last use of the deadlift during training, or requiring LS subjects to have never learned or observed the deadlift, regardless of the level of instruction. However, because subjects in the LS group were recreationally active weight lifters, most with a history of athletic participation, and seemed to perform at least as well as the trained athletes in the HS group, the results of the study should be more applicable to college athletes on the Division I level. Comparison of our results with those from studies involving powerlifters, however, should take into account the difference in experience levels between the 2 populations. Lastly, only one camera was used, so no frontal plane data were collected, and comparison with studies using 3-dimensional analysis with multiple cameras should be done with caution. Future research should examine the training effects of the WID as well as measuring joint kinetics and kinematics in 3 dimensions.
Use of the walk-in deadlift can reduce low back muscle involvement compared with the conventional deadlift, but also shifts muscle involvement from the gluteus maximus to the vastus lateralis, highlighting the difference between the exercises. Given the reduction in low back muscle involvement in an exercise that poses some risk to the low back, it is recommended that strength professionals prescribe the WID for novice lifters or athletes recovering from injury to allow them to learn or relearn proper posture for the deadlift and other floor-based bar exercises in a lower risk situation. However, because most athletes will require experience with the CDL, either to perform the CDL itself effectively or as the basis for more advanced powerlifts such as the power clean and the snatch, practitioners should guide athletes towards the CDL as soon as is safe and practical.
1. Bird S, Barrington-Higgs B. Exploring the deadlift. Strength Cond J 32: 46–51, 2010.
2. Burden A. How should we normalize electromyograms obtained from healthy participants? What we have learned from over 25 years of research. J Electromyogr Kinesiol 20: 1023–1035, 2010.
3. Calhoon G, Fry AC. Injury rates and profiles of elite competitive weightlifters. J Athl Train 34: 232–238, 1999.
4. Camara KD, Coburn JW, Dunnick DD, Brown LE, Galpin AJ, Costa PB. An examination of muscle activation and power characteristics while performing the deadlift exercise with straight and hexagonal barbells. J.Strength Cond Res 30: 1183–1188, 2016.
5. Cholewicki J, McGill SM, Norman RW. Lumbar spine loads during the lifting of extremely heavy weights. Med Sci Sports Exerc 23: 1179–1186, 1991.
6. Cohen J. Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Hillsdale, NJ: Lawerence Earlbaum Associates, 1988.
7. Escamilla RF, Francisco AC, Fleisig GS, Barrentine SW, Welch CM, Kayes AV, Speer KP, Andrews JR. A three-dimensional biomechanical analysis of sumo and conventional style deadlifts. Med Sci Sports Exerc 32: 1265–1275, 2000.
8. Escamilla RF, Lowry TM, Osbahr DC, Speer KP. Biomechanical analysis of the deadlift during the 1999 Special Olympics World Games. Med Sci Sports Exerc 33: 1345–1353, 2001.
9. Finucane SD, Rafeei T, Kues J, Lamb RL, Mayhew TP. Reproducibility of electromyographic recordings of submaximal concentric and eccentric muscle contractions in humans. Electroencephalogr Clin Neurophysiol 109: 290–296, 1998.
10. Gullett JC, Tillman MD, Gutierrez GM, Chow JW. A biomechanical comparison of back and front squats in healthy trained individuals. J Strength Cond Res 23: 284–292, 2009.
11. Haff GG, Triplett NT. Essentials of Strength Training and Conditioning (4th ed.). Champaign, IL: Human Kinetics, 2016.
12. Hales ME, Johnson BF, Johnson JT. Kinematic analysis of the powerlifting style squat and the conventional deadlift during competition: Is there a cross-over effect between lifts? J Strength Cond Res 23: 2574–2580, 2009.
13. Hermens HJ, Freriks B, Disselhorst-Klug C, Rau G. Development of recommendations for SEMG sensors and sensor placement procedures. J Electromyogr Kinesiol 10: 361–374, 2000.
14. Keogh J, Hume PA, Pearson S. Retrospective injury epidemiology of one hundred one competitive oceania power lifters: The effects of age, body mass, competitive standard, and gender. J Strength Cond Res 20: 672–681, 2006.
15. McGuigan MRM, Wilson BD. Biomechanical analysis of the deadlift. J Strength Cond Res 10: 250–255, 1996.
16. Swinton PA, Stewart A, Agouris I, Keogh JW, Lloyd R. A biomechanical analysis of straight and hexagonal barbell deadlifts using submaximal loads. J Strength Cond Res 25: 2000–2009, 2011.
17. Winwood PW, Hume PA, Cronin JB, Keogh JW. Retrospective injury epidemiology of strongman athletes. J Strength Cond Res 28: 28–42, 2014.
18. Yavuz HU, Erdag D, Amca AM, Aritan S. Kinematic and EMG activities during front and back squat variations in maximum loads. J Sports Sci 33: 1058–1066, 2015.