Objective. To compare two methods of calibrating the erector spinae electro
myographic signal against moment generation in order to predict extensor mo
ments during asymmetric lifting tasks, and to compare the predicted moments
with those obtained using a linked-segment model.
Methods. Eight men lifted loads of 6.7 and 15.7 kg at two speeds, in varyin
g amounts of trunk rotation. For each lift, the following were recorded at
60 Hz; the rectified and averaged surface electromyographic signal. bilater
ally at T10 and L3, lumbar curvature using the 3-Space Isotrak. movement of
body segments using a 4-camera Vicon system, and ground reaction forces us
ing a Kistler force-plate. Electromyographic (EMG) and Isotrak data were us
ed to calculate lumbosacral extensor moments using the electromyographic mo
del, whereas movement analysis data and ground reaction forces were used to
estimate net moments using the linked-segment model. For the electromyogra
phic technique, predictions of extensor moment were based on two different
sets of EMG-extensor moment calibrations: one performed in pure sagittal fl
exion and the other in flexion combined with 45 degrees of trunk rotation.
Results. Extensor moments predicted by the electromyographic technique incr
eased significantly with load and speed of lifting but were not influenced
by the method of calibration. These moments were 7-40% greater than the net
moments obtained with the linked-segment model, the difference increasing
with load and speed.
Conclusions. The calibration method does not influence extensor moments pre
dicted by the electromyographic technique in asymmetric lifting, suggesting
that simple, sagittal-plane calibrations are adequate for this purpose. Di
fferences in predicted moments between the electromyographic technique and
linked-segment model may be partly due to different anthropometric assumpti
ons and different amounts of smoothing and filtering in the two models, and
partly due to antagonistic muscle forces, the effects of which cannot be m
easured by linked-segment models.