Trunk muscle activation and associated lumbar spine joint shear forces under different levels of external forward force applied to the trunk

1. Introduction 

Fracturing of the endplate due to excessive compression is thought to be a major cause of low back pain incidence [33]. Consequently, guidelines in ergonomics, aimed at reducing the risk of developing low back pain, have mainly been based on limiting compressive loading [34] and a main focus of low back research has been quantification of compression forces at the spine [2], [11], [22], [23], [31], [32].

The importance of shear forces on the human spine is still a matter of debate. Large ranges of estimated joint shear forces and shear strength can be found in the literature. Failure of the pars interarticularis has been shown to occur at joint shear forces ranging from about 600–4000N [6], [15]. With respect to the joint shear loading in daily life, estimates of the joint shear forces at the L4L5 joint during lifting have been reported to be relatively low (i.e., below 300N) for a range of load weights [24], load positions [13] and lifting techniques [13]. At the L5S1 joint, joint shear forces of much higher magnitude (i.e., between 600 and 1500N) have been reported during lifting [8], [13], [14], [19]. Joint shear forces of such magnitudes might exceed spinal (intervertebral) shear tolerance. Lumbar spine joints are crossed by a large number of muscle fascicles, with different lines of action with respect to the spine [16], [17]. The resulting mathematical indeterminacy of the lumbar system (i.e. muscles and joints) might, theoretically, allow the central nervous system to recruit lumbar muscles in such a way, that shear forces are limited, or even minimized [10], [21]. For instance, a more backward oriented average muscle force could be obtained by shifting abdominal muscle activation from the external oblique to the internal oblique and by shifting back muscle activation from the multifidus to the longissimus. Potentially, such a strategy might help to reduce the risk of shear strain overloading of the intervertebral joint.

The question whether the neuromuscular system applies such a strategy can be addressed by varying externally applied forward forces on the trunk while keeping the moment that the muscles have to generate, constant [1]. In the current study, a constant moment loading relative to the L3L4 joint was maintained, while examining the effect on the neuromuscular system of an increase in externally applied forward forces at the trunk. Muscle activation and spinal curvature were measured, and resulting shear and compression forces at all lumbar intervertebral joint levels were calculated. It was hypothesized that, under constant moment loading relative to L3L4, an increase in external forward force on the trunk results in a shift in muscle activation towards the use of muscles that have a line of action pulling the trunk more backward, thereby reducing the increase in total joint shear force (i.e, the shear force due to external forces plus muscle forces plus gravity; see Fig. 1).

View Large Image Download to PowerPoint

Fig. 1. (A)–(C) Free body diagrams illustrating the hypothesis in this study. Fext is the external force, applied to the subject, mg is the force on the upper body due to gravity, which is assumed to have no moment arm relative to the intervertebral discs. Fmusc is the muscular force in response to Fext. Note that Fmusc (with an average muscle pull angle α relative to transversal surface through an intervertebral disc) represents the summed forces of back muscles and abdominal muscles. The reaction force at the trunk, here indicated at the L3L4 joint, is decomposed into a shear component (Fr,shear) and a compression component (Fr,comp) being perpendicular and parallel to the upper endplate of L4 (indicated by dotted lines), respectively. When the small Fext (A) is replaced by a larger Fext that is applied at a lower point on the trunk the required muscular moment remains roughly unchanged but Fr,shear increases (B). It is hypothesized that the neuromuscular system responds to this increased shear force challenge by using trunk muscles with a more backward directed line of action, in order to reduce or limit Fr,shear (C). (D) and (E) are illustrations (not free body diagrams) showing how both a shift in activation of abdominal muscles (from Fabd,f to Fabd,b) and a shift in activation of back muscles (from Fback,f to Fback,b) could affect the direction of Fmusc. Note that in the free body diagrams Fshear and Fr,comp are indicated relative to the trunk. However, to be consistent with the literature, calculated shear forces and compression force will be reported relative to the pelvis, thus having an opposite direction. Relative to the pelvis, forward shear forces and downward compression forces are reported as having a positive sign. To facilitate interpretation of the data, Fr,shear and Fr,comp were decomposed in a component due to Fmusc and a component due to Fext plus mg. This decomposition is not shown in the free body diagrams.

2. Methods 

2.1. Participants and procedure 

Twelve male participants (age 24 (SD=2) yrs, weight 73.8 (SD=9.5) kg, height 1.79 (SD=7.4) m) participated in the experiment after signing an informed consent. Participants isometrically resisted forward forces at the trunk, while the lumbar curvature and the activity of superficial trunk muscles were measured. The forward forces, of about 5s each, were applied at two moment load magnitudes (30 and 50Nm) relative to the L3L4 joint. For each moment magnitude, three heights (the spinous processes of T3, T6 and T9) were used to position the forward force (Fig. 2). Relative to the L3L4 joint, the lever arms were 170 (SD=13) mm (T9), 261 (SD=14) mm (T6) and 352 (SD=16) mm (T3). Force magnitudes were, for each participant individually, adapted to those lever arms, to obtain the desired moment. As a result, the (externally applied) forward force was more than twice as large at T9 compared to T3, without affecting the net moment at the L3L4 joint (see Fig. 2A). Note however that in this experimental setup moments can only be held constant relative to one point so that, at other lumbar joints, moments do vary with the lever arm. The order of moment and forward force conditions was randomized. Forces were applied in a purely horizontal direction, using a 30×115mm padded surface that had been attached to a sliding arm via a hinge joint (Fig. 2B). The padded surface prevented discomfort, even at the highest force magnitudes (294N, SD=21N). During the exertions, the pelvis of the participants was strapped (with two straps) to a frame, and participants were standing on a platform with wheels, that caused high friction in sideward direction but negligible friction in forward–backward direction. The purpose of the platform was to minimize net moments at the hip joints and thereby to minimize the contribution of muscles that (besides crossing the hip joint) might produce shear forces at the lumbar spine, but are not accessible for surface EMG measurement. In the absence of horizontal forces under the feet, the support at the pelvis provided the backward force that was needed to prevent forward translation of the participant and the support plus straps provided the backward moment needed to prevent forward rotation of the participants around the pelvic support (Fig. 2D).

View Large Image Download to PowerPoint

Fig. 2. A graphical representation of the experimental setup. Participants were standing on a rolling platform with their pelvis strapped (with double straps) to a frame while horizontal forces were applied to the trunk with a padded surface (A). Force magnitude and location at the trunk were varied in such a way that a constant moment relative to L3L4 was obtained (B). Participants received real-time feedback of either one marker (at C7) or two markers (at C7 and L3) and they were asked to position the markers equal to the positions that had been recorded in a neutral posture without an external load (C). The rolling platform prevented horizontal forces under the feet. A pelvic support supplied a backward force (Fsup) and a moment (Msup), needed to prevent forward translation or rotation of the participant (D). Fg is the ground reaction force, Fext is the external force, applied to the subject, mg is the force on the whole body due to gravity.

3. Results 

Under equal moment loading relative to L3L4, the larger forward forces at T9 on the trunk resulted in significant changes in EMG amplitude in most trunk muscles (Table 1 and Fig. 3). For all intervertebral joints it was found that, compared to a small forward force at T3, a large forward force at T9 resulted in a slightly smaller estimated muscle pull angle (ranging from 4.4 (SD=3.1) at T12L1 to 0.8 (SD=0.8) degrees at L5S1; see Fig. 4). This is in line with the hypothesis, since it means a shift in muscle activation towards a more backward directed average muscular force with increasing externally applied forward forces on the trunk. However, the main finding was an overall increase in muscle activation rather than a shift in muscle activation. This increase was significant in four back muscles (ranging from 2.1 (SD=2.2) to 3.8 (SD=3.0) % MVC for the 50Nm moment magnitude) and in three abdominal muscles (ranging from 1.0 (SD=0.9) to 2.1 (SD=1.5) % MVC for the 50Nm moment magnitude). As a result, model calculations showed, with larger (external) forward forces at T9, a significantly larger magnitude of the estimated muscular shear force (at 50Nm moment the increase ranged from 23.5N, (SD=25.9N) at L4L5 to 102.1 (SD=104.0) N at L5S1) and compression forces (at 50Nm moment the increase ranged from 208.1 (SD=190.7) N at T12L1 to 324 .1 (SD=254.0) N at L5S1) in all intervertebral joints (Fig. 5, Fig. 6, Table 1). The direction of the estimated muscular shear force (relative to the upper surface of the lower of each pair of vertebrae) was backward for the T12L1 to the L4L5 joints, but forward (and larger in magnitude) for the L5S1 joint. Consequently, the overall increase in muscle activation tended to reduce the resulting forward shear force at the T12L1 to L4L5 joints (at 50Nm moment down to on average below 200N for Th12L1 to L3L4 and down to 436.2 (SD=63.9) N for L4L5). At the same time, the muscular response tended, despite the 0.8° more backward muscle pull angle, to increase the resulting shear force at the L5S1 joint (at 50Nm moment up to 1080.1 (SD=150.4) N). The latter finding contradicts the hypothesis.

Table 1. Results of repeated measures ANOVA’s

		FB	MOM	POS	FBxMOM	FBxPOS	MOMxPOS	FBxMOMxPOS
		p-value	p-value	p-value	p-value	p-value	p-value	p-value
	Rectus abdominis	0.461	0.039	0.083	0.698	0.363	0.085	0.413
	Obl. int. lat.	0.194	0.014	<0.001	0.856	0.221	0.052	0.241
	Obl. ext. lat.	0.748	0.001	0.002	0.661	0.603	0.050	0.555
	Obl. int. ant.	0.829	0.111	0.549	0.191	0.150	0.306	0.214
	Obl. ext. dors.	0.321	0.001	0.004	0.129	0.713	0.021	0.642
	Longissimus thor.	0.766	<0.001	0.056	0.746	0.438	0.048	0.562
	Iliocostalis thor.	0.486	0.012	0.100	0.178	0.388	0.224	0.275
	Latissimus dorsi.	0.186	0.005	0.006	0.586	0.456	0.012	0.772
	Iliocostalis lumb.	0.067	<0.001	0.005	0.075	0.743	0.046	0.542
	Longissimus lumb.	0.497	<0.001	0.008	0.156	0.826	0.231	0.440
	Multifidus	0.550	<0.001	<0.001	0.042	0.896	0.743	0.502
	Total shear T12L1	0.336	<0.001	<0.001	0.946	0.815	0.405	0.135
	L1L2	0.421	0.001	<0.001	0.748	0.830	0.374	0.146
	L2L3	0.514	0.009	<0.001	0.511	0.835	0.311	0.165
	L3L4	0.849	0.001	<0.001	0.354	0.534	0.063	0.252
	L4L5	0.990	<0.001	<0.001	0.912	0.318	<0.001	0.237
	L5S1	0.956	<0.001	<0.001	0.023	0.677	0.020	0.533
	Musc. shear T12L1	0.588	0.005	0.001	0.519	0.692	0.007	0.066
	L1L2	0.688	0.001	<0.001	0.381	0.741	0.007	0.065
	L2L3	0.830	<0.001	<0.001	0.231	0.818	0.008	0.068
	L3L4	0.783	<0.001	<0.001	0.133	0.915	0.014	0.089
	L4L5	0.422	0.023	0.016	0.358	0.615	0.013	0.129
	L5S1	0.882	<0.001	0.001	0.032	0.693	0.175	0.603
	Total comp T12L1	0.480	<0.001	<0.001	0.066	0.807	0.093	0.582
	L1L2	0.528	<0.001	<0.001	0.055	0.823	0.103	0.595
	L2L3	0.594	<0.001	<0.001	0.046	0.831	0.113	0.611
	L3L4	0.618	<0.001	<0.001	0.037	0.864	0.136	0.644
	L4L5	0.668	<0.001	<0.001	0.038	0.858	0.178	0.641
	L5S1	0.565	<0.001	0.001	0.040	0.881	0.278	0.659
	Musc. comp T12L1	0.523	<0.001	<0.001	0.052	0.854	0.095	0.523
	L1L2	0.564	<0.001	<0.001	0.045	0.869	0.105	0.540
	L2L3	0.631	<0.001	<0.001	0.037	0.885	0.111	0.555
	L3L4	0.655	<0.001	<0.001	0.029	0.904	0.114	0.581
	L4L5	0.730	<0.001	<0.001	0.028	0.911	0.119	0.564
	L5S1	0.691	<0.001	<0.001	0.025	0.956	0.126	0.559
	Musc. pull angle T12L1	0.561	0.004	<0.001	0.371	0.607	0.018	0.561
	L1L2	0.763	0.005	<0.001	0.370	0.654	0.015	0.763
	L2L3	0.986	0.006	<0.001	0.353	0.743	0.015	0.986
	L3L4	0.391	0.015	0.003	0.429	0.836	0.014	0.391
	L4L5	0.286	0.042	0.020	0.502	0.671	0.025	0.286
	L5S1	0.184	0.239	0.004	0.619	0.489	0.486	0.184
	Lumbar angle	0.002	0.001	<0.001	0.778	0.581	0.569	0.059

Independent variables were number of feedback points (FB), moment level (MOM) and position of forward force on the trunk (POS). Dependent variables were activation of trunk muscles, muscular as well as total shear force and compression (comp) force, and average muscle pull angle at all lumbar intervertebral joints. Bold numbers indicate significant p-values (p<0.050).

View Large Image Download to PowerPoint

Fig. 3. Average muscle activation expressed as a percentage of MVC of five abdominal and six back muscles during isometric exertions at two different moment magnitudes (30 and 50Nm relative to the L3L4 joint). At each moment magnitude a high forward force was applied at a low position on the thorax (T9), an intermediate force was applied at a medium position on the thorax (T6) and a low force was applied at a high position on the thorax (T3). Error bars indicate 1 standard error of the mean. Data were averaged over two feedback conditions (C7 or C7 and L3 position).

View Large Image Download to PowerPoint

Fig. 4. The average muscle pull angle estimated at six intervertebral joints by an EMG-assisted model during isometric exertions at two different moment magnitudes (30 and 50Nm relative to the L3L4 joint). At each moment magnitude a high forward force was applied at a low position on the thorax (T9), an intermediate force was applied at a medium position on the thorax (T6) and a low force was applied at a high position on the thorax (T3). Error bars indicate 1 standard error of the mean. Data were averaged over two feedback conditions (C7 or C7 and L3 position).

View Large Image Download to PowerPoint

Fig. 5. Shear forces, estimated at six intervertebral joints by an EMG-assisted model during isometric exertions at two different moment magnitudes (30 and 50Nm relative to the L3L4 joint). At each moment magnitude a high forward force was applied at a low position on the thorax (T9), an intermediate force was applied at a medium position on the thorax (T6) and a low force was applied at a high position on the thorax (T3). For each joint, total joint shear forces are indicated. In addition, the total joint shear force is decomposed in a component caused by external force plus gravity and a component caused by muscle forces. Positive forces are anterior forces on the pelvis. Error bars indicate 1 standard deviation. Data were averaged over two feedback conditions (C7 or C7 and L3 position).

View Large Image Download to PowerPoint

Fig. 6. Compression forces, estimated at six intervertebral joints by an EMG-assisted model during isometric exertions at two different moment magnitudes (30 and 50Nm relative to the L3L4 joint). At each moment magnitude a high forward force was applied at a low position on the thorax (T9), an intermediate force was applied at a medium position on the thorax (T6) and a low force was applied at a high position on the thorax (T3). For each joint, total joint compression forces are indicated. In addition, the total joint compression force is decomposed in a component caused by external force plus gravity and a component caused by muscle forces. Error bars indicate 1 standard deviation. Data were averaged over two feedback conditions (C7 or C7 and L3 position).

The analyses of the lumbar curvature showed that larger (externally applied) forward forces at T9 resulted in a more flexed posture, i.e., in a reduced lumbar lordosis (Fig. 7 and Table 1).

View Large Image Download to PowerPoint

Fig. 7. Average lumbar curvature (relative to neutral model posture) adopted during isometric exertions at two moment magnitudes (30 and 50Nm) and with one and two point feedback with respect to the trunk posture. At each moment magnitude a high forward force was applied at a low position on the thorax (T9), an intermediate force was applied at a medium position on the thorax (T6) and a low force was applied at a high position on the thorax (T3). Error bars indicate 1 standard error of the mean.

As might be expected, a 50Nm moment magnitude resulted in a higher activation in almost all muscles and in larger compression and shear force estimates at all intervertebral joints than a 30Nm moment magnitude (Fig. 3, Fig. 5, Fig. 6; Table 1). In addition, the lumbar lordosis was reduced with larger moments and the muscle pull angle showed a more backward directed average muscle force.

Furthermore, providing feedback from the C7 and the L3 marker rather than from the C7 marker alone did not prevent an increase in lumbar flexion with increasing shear force. The number of feedback points had a main effect on lumbar flexion but not on muscle activation, shear forces, compression forces, or the muscle pull angle. However, there were some negligible interactions with moment magnitude (see Table 1).

4. Discussion 

Consistent with shear force calculations during lifting [13], the current results showed that, in upright standing posture, joint shear force estimates are highly dependent on the vertebral level that is studied. This is mainly the consequence of the orientation of the vertebrae. In neutral upright standing posture, the L5S1 joint is around 30° more forward inclined than the L4L5 joint. Since many muscle slips span both joints, a much more forward directed muscular force at the L5S1 joint is thus not unexpected.

The hypothesis in this study was that, under constant moment loading relative to L3L4, an increase in external forward force on the trunk results in a shift in muscle activation towards the use of muscles that have a line of action pulling the trunk more backward, thereby reducing the increase in total shear force at the lumbar intervertebral joints. The muscle pull angle indeed suggested a small (though significant) change towards a more backward directed average muscle line of action, being in part due to a flattening of the lumbar lordosis [21]. However, the main response of the neuromuscular system to increasing shear force challenges appeared to be an overall increase in muscle activation, as evidenced by an increased activation in most trunk muscles. This increase was small in terms of % MVC (maximally 3.8%). However, considering the overall level of activation of the back muscle (on average below 27% MVC in all muscles and all conditions), the relative increase was quite substantial. As a result, a substantial overall increase in estimated muscular compression as well as shear forces was seen with larger (externally applied) forward forces. As the muscular shear force was backward at the L4L5 up to T12L1 joint level but forward at L5S1, the increased muscle force caused a reduction of the forward total joint shear forces at L4L5 up to T12L1 but an increase in forward total joint shear force at L5S1. Notably, the forward total joint shear force at L5S1 was already much higher than at the other joints. The overall increase in muscle activation increased the muscular shear force further with on average 24%. This is (despite a flattening of the lumbar spine) more than the increase in moment relative to L5S1 (note that moments were held constant at L3L4, but increased with increasing forward force by on average 16% at L5S1).

Consequently, it is concluded that the hypothesis was not supported by the data. Rather than a specific action aiming to counteract shear forces, the neuromuscular system appeared to respond with an overall increase in muscle activity. This response contributes to a general stiffening of the spine [5]. In addition, increased abdominal muscle activation increases the intra-abdominal pressure, which likely provides additional spine stability [3], [4] and enhances the resistance against shear in intervertebral joints [12]. However, at the L5S1 joint it is unclear, whether such an increased resistance against shear forces outweighs the increase in muscular forward shear force that is associated with the overall increase in trunk muscle activation. Note that shear injury (spondylolisthesis) is most common at the L5S1 joint [18].

In contrast to the current study, Callaghan and McGill [1] found more antagonistic co-contraction in pure moment loading than in moment plus forward force loading. However, in that study loads were applied via the arms, causing the pure moment loading to require high moments at the shoulder joint. The strong activity of the shoulder muscles may have caused a high activity of the trunk muscles to stabilize the thorax.

It could be speculated that the need to obtain moment equilibrium simultaneously at all intervertebral joints rather than the shear load at the joints caused the overall increase in muscle activation in the current study. The moment difference between the subsequent lumbar intervertebral joint levels was largest in the short moment arm/high forward force conditions. This may have required additional co-contraction due to the difficulty in reaching moment equilibrium in all intervertebral levels simultaneously. The absence of an effect of forward forces, applied directly at the L4L5 joint [25] supports this explanation.

In the short moment arm/high forward force conditions, we also found a flattening of the lumbar spine. Most likely, this occurred through pelvic extension, despite tight strapping of the pelvis. This flattening, just like the additional co-contraction, may have been caused by the difficulty to obtain moment equilibrium in all intervertebral levels simultaneously. Conversely, the flattening may have been allowed by the neuromuscular system as a strategy to shift the load from the posterior elements to the intervertebral disc. This is consistent with the shear injury threshold, which is likely to be larger for the intervertebral disc than for the posterior elements [36]. The effect of moment arm/forward force was, in the two-point feedback condition, comparable to the effect in the one-point feedback condition, but the lumbar spine was overall flatter in the two-point feedback condition. This supports the non-intentional character of the flatter back, because controlling two feedback points is substantially more difficult (thus requiring attention to be directed at feedback points) for the subjects.

A limitation of the current study is that shear force estimates are highly dependent on assumptions regarding the lines of action of the muscles relative to the vertebrae [31]. It should also be realized that we only measured the EMG of superficial back muscles, and that only 11 EMG signals (representing 22 muscles under symmetry assumptions) were used to drive 130 muscle slips. The question therefore arises whether the current EMG-driven model would be able to detect changes in shear forces from changes in muscle activation. To explore the upper and lower bounds of shear forces that could result from changes in muscle activation inserted in the EMG-driven model, we calculated (for the L5S1 joint) the ratio of the shear force to extensor moment separately for each of the 11 muscles (i.e., groups of muscle slips). After averaging over conditions and participants those ratios were found to range from 4.2 to 9.4 for the back muscles, and from 23 to −53 for the abdominal muscles. This indicates that a redistribution of activation over trunk muscles could have resulted in substantial changes of muscular shear forces without necessarily affecting moments.

It has recently been shown that multifidus activation is not well-represented by surface EMG [27]. However, although it cannot be excluded that joint shear forces are actively counteracted by deep muscles that cannot be disclosed by the current approach, the shear component of the multifidus muscles is small [16], so that errors in measured multifidus EMG only have limited effects on estimated shear forces. In addition, the current results are based on a specific population (young healthy males), which limits the potential to generalize results. For instance, subject anthropometry has been shown to affect joint shear forces [7] and low back pain patients have been shown to have different contraction patterns in isometric trunk extension efforts [30]. Furthermore, it cannot be excluded that the shear loads on specific tissues rather than total joint shear forces trigger specific activation patterns [26]. The distribution of shear loads over ligaments, disc and facets, which likely depends on the posture, was not taken into account in this study. However, postural variations were also rather small. Finally, we realize that the experimental task differs from daily tasks. However, this task, holding an upright posture while horizontal forces are exerted on the trunk, enabled us to obtain a substantial variation in shear force challenge while maintaining maximum control over moments.

In summary, the current study showed that an increase of the applied forward forces at the trunk resulted in a flattening of the lumbar lordosis and an increase in co-contraction. This resulted in an increase of spinal compression, which enhances the stiffness of the lumbar spine [9]. No consistent evidence for a specific response to counteract intervertebral shear forces (i.e., a shift of activation to muscles resulting in a reduction of the increase in joint shear force) was found.