Chapter

Epilogue

This thesis started with two observations, emerging from recent epidemiologic reviews. One observation is that although many factors may contribute to the occurrence of low-back pain, physical loading is the factor that is most consistently found to be related to low-back pain. Aspects of physical loading that are regarded as the most important are lifting, whole body vibration and twisting and bending. The second observation was that more accurate estimation of mechanical loading consistently results in higher risk estimates for mechanical factors in relation to low-back pain. Consequently, it was pointed out that techniques used to study risk factors in manual materials handling activities, require improvements. The aims of this thesis were (1) to improve mechanical modeling in order to open up new perspectives for biomechanical research of lifting activities, (2) to investigate the effect of asymmetry in lifting movements on low-back loading and (3) to highlight some aspects of the tuning of the lifting movement to the object to be lifted.

limitations of the linked segment model approach and challenges for future research

There are several limitations to the use of linked segment models in the study of low-back loading. One of them is that it only results in an estimate of net reactive moments and forces at joints. The actual loading in the joint is not calculated since antagonistic contractions are not taken into account and the distribution of the torque over muscles and between active and passive structures is not known. To obtain estimates of internal loading, the output of a linked segment model could be used as input for muscle models that estimate this loading with the aid of a detailed representation of the trunk anatomy combined with criteria to predict the distribution of activation between muscles (Dieën, 1997). Such an approach is necessary for example if one wants to relate the actual loading to the strength of structures in the spine. A disadvantage of such a step is that the uncertainty of the outcome increases, especially due to the problem that it is not known what criteria the central nervous system actually uses to distribute activity over muscles. Thus, a challenge for future research is to find muscle activation distribution criteria that represent the functioning of the central nervous system as good as possible. Alternatively, EMG-measurements could be used to estimate the actual distribution of loading over muscles (Looze et al., submitted; Marras and Sommerich, 1991; McGill and Norman, 1986). Recently, hybrid approaches, combining EMG and optimization, were developed (Cholewicki et al., 1995; Dieën and Looze, accepted). Coupling of such models to a 3-D linked segment model seems the next step to be taken.

Another limitation of linked segment models is that they can not be used to estimate effects of repeated or sustained mechanical loading on the risk of low-back pain. Additional measurements of muscle fatigue (Dieën et al., 1995) or the use of models to predict the effect of repeated or sustained loading on spinal strength (Dieën and Toussaint, 1995) may be required to incorporate time dependent effects of mechanical loading of the lumbar spine. Another option is quantification of fluid loss from the intervertebral disk. Fluid loss from the intervertebral disk is thought to play an important role in the increase of the risk of damage due to repeated or sustained loading, through the development of annular stress peaks with fluid loss (Adams et al., 1996). Indirect indications of fluid loss from the intervertebral disks can be obtained by measurement of spinal shrinkage (Dieën and Toussaint, 1993). Recently, a new method, combining finite element technology and MRI measurements, was developed to obtain a detailed picture of fluid loss from and fluid redistribution in the intervertebral disc (Kingma et al., 1998).

model developments: new perspectives for biomechanical research of lifting

Regarding the first aim of this thesis, the results in chapter 2 were somewhat disappointing. With a more detailed anthropometric model improvements were found for back-lifting movements but not for leg-lifting movements. It can not be exluded that the use of another anthropemetric model, either of a geometrical (e.g. Hanavan, 1964 or Hatze, 1980) or proportional (e.g. Zatsiorsky and Seluyanov, 1985) nature, would give better results. In addition, division of the trunk into more than one segment might improve the top-down calculation of net torques at the lumbosacral joint.

Chapter 3 showed that substantial improvements can be obtained for estimates of the body COM using an optimization procedure. Errors are reduced to a level that allows relating the COM to the ground reaction force vector. At the time of finishing this thesis, the method proposed in chapter 3 has already been applied successfully in several studies concerning human balance during lifting (Commissaris and Toussaint, 1997a; Toussaint et al., 1997a; Toussaint et al., 1995a; Toussaint et al., 1997b). A major limitation of the method proposed in chapter 3 is that the improvements are specific for the COM trajectory. It does not guarantee improvements in estimates of joint loadings.

The validation of the 3-D linked segment model in chapter 4 showed satisfactory results. As a consequence, and in contrast to studies published to date, it was possible to obtain reliable estimates of the lateral flexing and twisting torques at the lumbo-sacral joint in asymmetrical lifting movements (chapter 5). It was shown that, even with 10^o of asymmetry in a lifting movement, subjects do not tend to twist their pelvis far enough to prevent asymmetrical low-back loading. In fact, pelvic twist accounted more or less constantly for about 25 % of the total lifting asymmetry. Asymmetric loading increases the internal (compressive) loading of the interverterbral disc through the need of co-contractions (Marras and Mirka, 1992). In addition, if there is also asymmetric trunk movement, the compressive load is likely to be distributed over a reduced number of annular fibers, resulting in increased local fiber strain (Shirazi-Adl, 1989). Thus, asymmetric loading of the lumbar spine, which consistently increases with increasing lifting asymmetry, might be an important contributing factor to the development of low-back pain. This is consistent with epidemiologic findings (Kelsey et al., 1984; Marras et al., 1995). It may be concluded from this thesis that it is desirable as well as possible to incorporate quantification of asymmetric lumbar loading in future epidemiologic and ergonomics research. Besides that, it was shown in chapter 6 that a sagittal plane analysis causes errors in the estimated lumbo-sacral extending torque of about 20 % if a lifting movements is 30^o asymmetrical. This further underlines the need for a full 3-D analysis if, as often will be the case in occupational lifting, a lifting movement is not completely symmetrical. Furthermore, even in symmetrical lifting it can be important to quantify asymmetrical movement or torque components since fatigue results in increasing asymmetry in lifting movements that are intended to be symmetrical (Dieën et al., 1998).

It is realized that a 3-D analysis is more complex than a 2-D analysis and thus less well-suited for large-scale epidemiologic research. However, taking recent improvements in 3-D movement analysis systems into consideration, the number of potential applications of a full 3-D analysis is increasing, especially in ergonomics research. It may often be possible to use a 2-step approach, starting with identification of possible hazardous working situations, followed by a detailed mechanical analysis of these situations and of possible alternatives, with a limited number of subjects. Such an approach was successfully applied in a 2-D mechanical analysis of the working situation of nurses (Looze et al., 1994b), brick-layers (Looze et al., 1996) and refuse collectors (Looze et al., 1995). For the time being, application of the current 3-D model to epidemiologic research will be limited to validation of instruments that allow large scale monitoring of low-back loading in occupational lifting. Currently, the results of the current model are compared to two systems, based on EMG measurement, trunk motion measurement and (for one system) neural networking, and designed for large scale research in working environments (Baten et al., 1996; Dolan et al., accepted). Hopefully, this will result in an improved insight in the dose-response relationship between physical loading and low-back pain.

It should also be pointed out that, although the 3-D linked segment model that was developed in this thesis, was applied to lifting movements, there is no reason to restrict its use to lifting movements only. Evidently, the model is a general purpose 3-D linked segment model. The model or parts of it could be applied in many areas of biomechanical research.

tuning the lifting movement to the object to be lifted

The microgravity experiment that was described in chapter 7 and 8, focused on two phenomena (the size-weight illusion and COM control respectively) affecting the stability of the subject during lifting as well as the loading of the lumbo-sacral joint.

In chapter 7 it was suggested that the persisting elevated initial effort (and the associated size-weight illusion) in repeatedly lifting two boxes of equal mass but different volume, may be related a persisting upward scaling of the force component necessary to accelerate the object rather than the force component necessary to overcome the weight of the object. This could indicate that conscious knowledge of object mass as well as experience in previous lifts results in adaption of the weight-related lifting force but not in adaptation of the force, necessary to accelerate the object.

In chapter 8 it was shown that control of the horizontal COM position has at least two functions. One of them, the control of equilibrium, is well known. Under microgravity this function is no longer relevant. The other function is, together with the ground reaction force vector, to generate an external moment, adequate for the required rotational body movement. This function is still important under microgravity and may be the reason for the backward displacement of the COM in movements involving trunk bending under microgravity. Still, it must be realized that the plausibility of these functions, and the findings in the microgravity experiment, do not prove that the COM is the parameter that is actually being controlled by the central nervous system.

The results from chapter 7 and 8 do suggest that a biomechanical analysis can be fruitful in research on questions regarding motor control and perception. Evidently, chapter 7 and 8 can be regarded only as a small contribution to the research in this field.