Chapter 1

Introduction

Low-back pain has been recognized as a major cause of sickness and disability for several decades. Every year, 2 to 5 % of the population receives medical care or loses time from work as a consequence of low-back pain (Frank et al., 1996). In 1989 about 33 % of the costs of worker’s compensation claims in the USA resulted from low-back pain (Webster and Snook, 1994). Consequently, low-back pain is estimated to be the most costly ailment of working-age adults (Frank et al., 1996). About three quarters of the direct costs (medical costs and compensation for lost time) is caused by a small group (7.4 % of all claims resulting from low-back pain in Quebec, Canada) that is still absent from work after 6 months (Abenhaim and Suissa, 1987). Sickness claims due to low-back pain have increased over the past two decades in many industrialized countries (Frank et al., 1996). In addition, sick-leave due to low-back pain is not evenly distributed among professions. Absence from work due to low-back pain in the past two months was reported by 2.1 % of the workers in non-sedentary professions compared to 1.0 % in sedentary professions (Hildebrandt, 1995).

Risk factors for the development of low-back pain can be categorized in individual, psychosocial and physical factors (Winkel and Mathiassen, 1994). Epidemiologic studies indicate positive, negative as well as no associations for individual factors like gender and age (Ferguson and Marras, 1997). Most epidemiologic studies failed to demonstrate any association of physical exercise or sport with prevention of low-back pain (Burdorf and Sorock, 1997). Positive associations with the risk of developing low-back pain are generally found for persons with previous episodes of low-back pain (Ferguson and Marras, 1997). The role of psychosocial factors in the risk of developing low-back pain is still unclear. In a prospective study among aircraft employees Bigos et al. (1991) found that subjects who stated that they "hardly ever" enjoyed their job tasks were 2.5 times more likely to report a back injury (P = 0.0001) than subjects who "almost always" enjoyed their job tasks. On the other hand, no association between physical load and low-back pain was found (Bigos et al., 1991). However, in recent reviews on epidemiologic risk factor research, these results were criticized because of the limited range of exposure to physical load (Frank et al., 1996), the small proportion of the low-back pain cases that could be predicted with the final statistical model in the study (Frank et al., 1996) and the methodology employed to measure the exposure to physical load (Burdorf and Sorock, 1997). In a thorough review on epidemiologic studies investigating psychosocial risk factors, Bongers et al. (1993) concluded that these studies "do not provide conclusive evidence due to the high correlation between psychosocial factors and physical load and to the difficulties in measuring dependent and independent variables". Still, it was also concluded that there was sufficient empirical evidence to relate monotonous work and time pressure to musculoskeletal symptoms (including low-back pain).

Winkel and Mathiassen (1994) pointed out that imprecise measurement of mechanical exposures may imply that the calculated relative risks are underestimated. If psychosocial factors correlate with mechanical exposure, and if these factors are measured with a higher precision, their risk ratio may be overestimated (Winkel and Mathiassen, 1994). This argument is further underlined in recent reviews on epidemiologic research. Ferguson (1997) reported that a positive association between low-back pain and physical loading is more often found when physical loading is measured directly. Burdorf and Sorock (1997) stated that "all of the studies with no apparent associations relied on exposure estimates derived from self-administered questionnaires, and none of these studies presented information on the reliability of these questions". In addition, NIOSH (1997) reported odds ratios between 1.2 and 5.2 for studies using subjective measures of exposure in lifting activities and odds ratios between 2.2 and 11 for studies that used more objective assessment of exposure.

When posture as well as duration of posture are quantified, several studies report dose-response relationships between working in awkward postures and the risk of developing low-back pain (Burdorf and Sorock, 1997). For instance, Punnet et al. (1991) showed increasing odds ratios for increasing duration of flexed postures, for increasing degrees of flexion and separate effects of lateral flexing/twisting. In all, there is a large amount of evidence that physical loading (i.e. lifting activities, whole body vibration and frequent twisting and bending) is a major risk factor for the development of low-back pain.

Determination of the relationship between work-related physical loading and low-back pain involves six basic steps. The first step is accurate measurement of posture and motion of the body and of external forces, applied to the body. The second step is quantification of the "external" load on the body, e.g. by using a mechanical model to calculate the (net) moment at the lumbar region, required to perform a (lifting) task. The third step is to determine what active (muscular) and passive structures contribute to this moment and whether there is any antagonistic activity. The fourth step is to determine the local stresses and strains in the tissues of the vertebral column, following from these forces. The fifth step is to relate these stresses and strains to the strength of the tissue in order to determine the "weak link" where damage might occur. The last step is to determine the consequences of damage, i.e. recovery, degeneration and/or pain. Evidently, accurate quantification of the net moment, required to perform a (lifting) task is necessary before reliable estimates in the next steps can be obtained. In addition, epidemiologic research, generally attempting to correlate "external" physical loading to low-back pain, might gain from more accurate moment calculation.

Since there are still several inaccuracies hindering appropriate quantification of physical loading, this thesis is mainly concerned with improvement of mechanical modeling in order to open up new perspectives for biomechanical research of lifting activities. In addition, as will be outlined later, some mechanical aspects of lifting that received little attention to date, will be investigated. These aspects concern asymmetric low-back loading in asymmetrical lifting and the tuning of the lifting movement to the object to be lifted.

The earliest attempt to (statically) quantify the mechanical loading of the low-back in lifting activities was made by Chaffin (1969). In this approach, the human body is regarded as a chain of rigid segments, interconnected by hinge joints. Using anthropometric measures, segment position data and external forces, Newtonian mechanics are applied to calculate reactive moments and forces at the joints. Leskinen et al. (1983a; 1983b) and Freivalds et al. (1984) incorporated the dynamics of movement into these so-called linked segment models. Looze et al. (1992b) introduced several improvements and provided a validation of the calculated reactive moments and forces at the lumbo-sacral joint by starting the calculations at both ends of the chain (i.e. at the hands and at the feet). The last model has successfully been applied in the evaluation of several aspects of lifting, like lifting technique (Dieën et al., 1994), the co-ordination of lifting (Toussaint et al., 1995b) and the difference between lifting and lowering (Looze et al., 1993). In addition, the model has been applied to assess the mechanical loading during different tasks in nurses (Looze et al., 1994b), brick-layers (Looze et al., 1996) and refuse collectors (Looze et al., 1995).

One aspect of lifting activities that has recently been brought to attention is the control of balance (Toussaint et al., 1995a). Equilibrium control in bipeds, having a relatively small base of support (i.e. the feet), is strongly challenged when a load is lifted, due to the sudden displacement of the center of mass (COM). Evidently, loss of balance during load handling can cause low-back pain and other injuries. In addition, attempts to maintain balance, especially in handling objects of unknown weight, could affect the mechanical loading of the low-back due to jerky movements. Whole body balance is completely determined by the relationship between the COM and the ground reaction force vector. One reason why balance related problems have received little attention might be that exploration of the relationship between the ground reaction force vector and the COM requires high precision measurement of the COM position. This high precision is necessary due to the fact that the moment arm between the ground reaction force vector and the COM is often no more than a few cm during lifting. Thus, errors in the prediction of the COM need to be quite small in order to obtain useful data. In chapter 2 and 3 of this thesis different approaches are used to improve COM estimates. In chapter 2 model performance during lifing movements in terms of whole body mechanics as well as lumbo-sacral mechanics for the Looze et al. (1992b) model are compared to the performance of a model with geometry based estimates of body segment mechanical properties. In chapter 3 an optimization approach is adopted to improve the estimation of the body COM trajectory during lifting.

Twisting movements were already mentioned as an important risk factor for the development of low-back pain. Kelsey et al. (1984) showed that twisting alone does not increase the risk of acute herniation of the intervertebral disc. Lifting loads of 11.3 kg or more only resulted in an increased risk if these lifting movements were performed more than 25 times a day. If such loads were lifted while twisting, a frequency below five times a day was enough to cause an increased risk of acute herniation. In addition, when a vertebral motion segment is loaded in vitro, compressive loading alone results in endplate fractures whereas compressive loading combined with flexion and twisting results in disc herniation (Gordon et al., 1991c).

Until recently, attempts to quantify asymmetric low-back loading in asymmetric lifting were not very successful. Kromodihardjo and Mital (1986; 1987) were the first authors who used a 3-D linked segment model to analyse (asymmetrical) lifting. However, their results were too variable to show a clear effect of asymmetry in lifting movements on the asymmetric components of low-back loading. Frigo (1990) studied symmetrical and asymmetrical lifting movements with a 3-D model but reported only results for the symmetrical lifting movements. Using a 3-D model, developed by Gagnon and Gagnon (1992), Plamondon et al. (1995) evaluated asymmetric lumbo-sacral loading in 0^o, 45^o and 90^o asymmetric lifting. However, they were not able to demonstrate lateral flexion or twisting moments significantly deviating from moments that were found during symmetrical lifting. Clearly, this is not consistent with epidemiologic findings nor with the intuitive feeling that such asymmetric moments should occur in asymmetric lifting. In addition, this is not consistent with findings reported from a simulation study on asymmetrical lifting (Jäger and Luttmann, 1992).

One explanation for the inconsistency could be that the models were not accurate enough to detect changes in asymmetrical moments due to increasing asymmetry in lifting movements. The main source of error in these models could be the use of skin markers on joint rotation centers. Such markers are known to move during joint motion due to movement of the skin around joints (Cappozzo et al., 1993). This may cause unstable local axis systems in body segments, resulting in large errors in asymmetrical moment components due to the erroneous projection of the major (extending) moment component on the lateral flexing and twisting axis.

Chapter 4 describes the development and validation of a full-body 3-D linked segment model that uses rigid cuffs on body segments instead of skin markers (Helm and Veeger, 1996) to follow body segment motions during lifting. Prior to the lifting movements the relationship between the cuff markers and anatomical landmarks is determined using a recording that can be termed "anatomical axis calibration" (Berme, 1980; Cappozzo, 1990).

In chapter 5, this model is used to quantify the asymmetrical low-back loading in lifting movements of increasing asymmetry. In addition, it is investigated to what extent subjects twist their pelvis to prevent asymmetric low-back loading.

Occupational lifting movements will mostly be asymmetrical to some extent. Application of 2-D models is simpler than application of 3-D models and might thus be preferred in occupational lifting research. The aim of chapter 6 is to get insight into the size of the errors that are made when a 2-D model is applied in asymmetrical lifting movements. This is achieved by simultaneous application of a 2-D and a 3-D linked segment model in symmetrical as well as asymmetrical lifting movements.

It could be questioned to what extent the mechanical loading of the low-back is influenced by the co-ordination of a lifting movement. The forward bending movement of the trunk as well as the addition of the mass of the object to be lifted cause a forward shift of the COM. This could result in a disturbance of whole body equilibrium, which is usually defined as maintaining the horizontal position of the body COM within the support area formed by the feet. It was shown that such a disturbance is counteracted in anticipatory fashion by an increased backward speed of the COM (Commissaris, 1997). In addition, initial lifting forces are scaled in a feed-forward way according to the expected mass of the object. Inadequate adaptation of the COM speed or of the initial lifting force, could result in (1) loss of balance and (2) increased low-back loading. In Chapter 7 and 8 it is attempted to obtain more insight in the factors that influence COM control and initial lifting forces by manipulating the effect of gravity. This was realized in an experiment that was made possible by the European Space Agency (ESA). Lifting movements were performed during so-called parabolic flights in a NASA KC-135 aircraft. During two flight days a total of 60 parabolas were executed. Within the aircraft the parabolic flight path causes brief microgravity periods so that the effect of gravitational forces is reduced to values close to zero. After a few minutes of steady flight, each parabola started with about 15 s of increased gravity (about 1.8 g) due to a fast pull-up of the aircraft. This period is followed by about 20 s of microgravity. Then, there is again 15 s of 1.8 g during which the aircraft pulls up to horizontal flight.

Chapter 7 focuses on a phenomenon that is generally overlooked in research on risk factors in occupational lifting (Luczak and Ge, 1989). This phenomenon, first described by Charpentier (1891), is called the size-weight illusion. It entails the phenomenon that for two objects of equal mass but different volume, subjects consistently report the larger object to feel lighter. Standards regarding maximal acceptable weights to be lifted (Waters et al., 1993), are generally based on weight perception (e.g. Garg and Badger, 1986; Mital and Fard, 1986; Mital and Manivasagan, 1983). Consequently, increasing maximum acceptable weights are sometimes reported with increasing object volume (Ciriello and Snook, 1983; Luczak and Ge, 1989). Yet, subjects use more force to lift larger objects (Gordon et al., 1991a; Gordon et al., 1991b). This is most likely due to the expectation of a higher force, required to lift the object. As a result, higher lumbar loading in lifting larger objects can be expected. In fact, lumbar loading during load pick-up is not determined by the actual weight of an object but by the expected weight (Commissaris and Toussaint, 1995). In chapter 7, the relative effects of mass and weight in the size-weight illusion are investigated. If subjects are able to largely eliminate the weight-related component of the lifting force under weightlessness, the elevated initial force in lifting a larger object would disappear in case it is exclusively due to overestimation of the force-component related to gravity (i.e. the weight of the object). If, on the other hand, the illusion is primarily caused by (expected) inertial properties the elevated initial lifting effort in the larger box should persist under microgravity.

Chapter 8 focuses on the regulation of balance during lifting. Under microgravity subjects do not fall when the COM travels out of the support plane formed by the feet. Thus, the necessity to control balance disappears. It is investigated if the control of COM movement during lifting movements remains comparable to normal gravity or if it is disturbed, functionally adapted to changed circumstances or if it completely disappears under microgravity.