Abstract
The authors studied the surface electromyographic (EMG) spectrum of the paraspinal muscles of 350 subjects. They were classified by their history as normal (n=175), chronic low back pain (n=145), or past history (n=30). They pulled upwards on a floor-mounted load cell at two-thirds of their maximum voluntary contraction for 30 s, while the EMG was measured from the paraspinal muscles at the L4/L5 level. From the EMG signal the root-mean-square (RMS) was calculated. Power spectrum analysis allowed calculation of the median frequency slope, the initial median frequency (IMF), modal frequency, peak amplitude and spectral width at half peak amplitude (half-width). All of the variables of the chronic group were significantly different from the normal group, except the median frequency slope, RMS slope and mode. Half-width, age and maximum voluntary contraction were shown to be independent predictors of back pain classification. Half-width classified the subjects with a sensitivity of 0.65 and a specificity of 0.75.