Background: Many outcome measures for the assessment of low back pain (LBP) have been described but few are specific or objective. Electromyography of the lumbar paraspinal muscles (EMG) has been studied for some years now but has been used only as a research tool, not clinically. Using maximum voluntary contraction (MVC) is convenient but this introduces subject volition and may reduce inter-subject comparability.

Methods: 491 subjects were tested. 209 were LBP subjects with 282 normal controls. For each the MVC was measured. EMG studies were performed only at 2/3 MVC in 367 subjects and a further 34 subjects were studied at a range of loads between 10% and 100% MVC. Loads were expressed as a proportion of the subjects lean body mass (LBM.)

Results: There were highly significant differences in MVC and load as a proportion of LBM between back pain subjects and controls. More than one in five back pain subjects were unable to manage loads of 0.5 LBM compared to 0.4% of controls. Less than 40% of back pain subjects could manage 1 LBM compared to almost 75% of normal subjects.

EMG variables were significantly influenced by load. For a 1LMB change in load most variables changed by at least 100%, notable exceptions being Half Width (27%) and Initial Median Frequency (IMF)(4%).

The ability of EMG variables to discriminate between normals and back pain subjects was examined in groups 0.1LBM wide. There were significant differences in Half Widths between the normal and back pain subjects in most groups, independent of load. There were no significant differences in Median Frequency Slopes (MFSlope) of normal and back pain subjects except at between loads of 1.4 to 1.5 LBM (p< 0.05). Accuracy of discrimination was poor, seldom better than 0.6 until subjects were tested at loads above1.2 LBM when accuracy increased sharply to 0.95 at loads between 1.4 to 1.5 LBM.

Conclusion: The usefulness of MFSlope as a discriminator has been variably reported in the literature. The present data indicate that in experimental conditions with subjects able to achieve loads greater than 1 LBM it is useful. However this load is not achievable in patients presenting clinically, and the Half Width, which is robust and reliable at low loads may be more valuable in the clinical setting.