Civil Engineering Reference
In-Depth Information
With errors between different calculation methods using the same data set approaching more than 100%
in some cases (Callaghan et al., 2001), it presents the problem that researchers will not be able to combine
exposure data from different studies or adopt a universal exposure TLV. Arguably, the absolute error in
cumulative load estimates might be of little concern when comparing relative exposure between two
groups (i.e., low back pain and controls). The assumption that error would be equal in the two
groups is also brought into question as error of these reduction approaches was influenced by the
type of task performed. The use of a reduced sampling rate (5 Hz approach) resulted in very small
errors across all tasks and subjects. This suggests that significant digitizing time might be saved
without compromising the accuracy of cumulative loading estimates. The sensitivity of some of the
approaches employed, in particular the square, work only and work
rest approaches, to the type of
task examined further brings into question the validity of using a single point in time to represent
cumulative exposure. Further reductions in sample rate were deemed to still produce accurate results
(Andrews and Callaghan, 2003). In fact, representing a task with kinetics calculated from data
sampled at 2 frames
/
sec only introduced an average of 3% error compared to kinetics at 60 frames
/
sec. While this represents a great reduction in the data required to document a task, processing
this volume of data is still quite labor-intensive. The reduction of sample rate to document
exposure has a further impact on model usage. As alluded to earlier in the discussion of dynamic
models, if data is reduced to a point where it does not sufficiently represent the frame by frame worker
movement, dynamic calculations of segmental accelerations cannot be performed or they produce erro-
neous values. The use of 2 frames
/
sec is well below the sample rates required for even slow moving
/
activities.
13.8.5 What Exposure Variable Should Be Used to Quantify Dose Exposure?
Mechanical loading of the spine has been used as a means to identify the possibility of developing a low
back injury for decades. Peak compression has formed the foundation for most ergonomic tools and
investigations of jobs with high rates of back injuries. Shear loading has more recently been identified
(Kumar, 1990; Marras and Granata, 1997; Norman et al., 1998) as another important biomechanical
factor to be considered when assessing jobs for the potential of back injuries. Both of these variables
have direct physiological relevance, although all variables from a biomechanical rigid link or joint
model also possess this relevance. One of the rationales for selecting a variable to be used is whether
it has a relationship to the injury or pain modes that it is trying to predict. While compressive
loading can definitely produce injuries, the primary injury (typically endplate failure) (Yingling et al.,
1997) may only lead to pain pathways through secondary developments in response to this injury. In con-
trast, repetitive flexion (Callaghan and McGill, 2001) or non-neutral compression (Gunning et al., 2001)
has the potential to damage both hard and soft tissues in the spine and joint moments may be a better
indicator of this mode of loading. Similarly, shear can also damage structures that have a much higher
density of pain sensing fibers such as the articular capsule or external margins of the intervertebral disc
(Yingling and McGill, 1999a). One of the strongest justifications at present for choosing a variable would
be the association between reporting of pain and the cumulative exposure that has been established in the
large-scale published epidemiology studies. These studies have primarily used cumulative compression
(Jager et al., 2000; Kumar, 1990; Norman et al., 1998; Seidler et al., 2001, 2003) with two reporting cumu-
lative anterior
posterior shear (Kumar, 1990; Norman et al., 1998) and only one examining cumulative
moment (Norman et al., 1998). A 2D-biomechanical model that incorporates both a rigid link model and
a lumbar joint model has the potential to examine five output variables where the equivalent approach in
3D yields nine kinetic variables (Table 13.2).
Reaction forces are output from a rigid link segment biomechanical model and are representative of
forces at the lumbar joint caused by body weight and the forces in the hands. Joint forces, also referred
to as “bone on bone” or net joint forces are calculated by using a joint model that incorporates muscle
forces and potentially passive tissue forces that are then combined with the reaction forces. Joint
/
Search WWH ::
Custom Search