Discriminant classification of motor unit potentials (MUPs) successfully separates neurogenic and myopathic conditions. A comparison of multi- and univariate diagnostical algorithms for MUP analysis.

Multivariate statistical methods may be more appropriate for the multidimensional material of quantitative motor unit potential (MUP) analysis than the multiple univariate tests of the conventional Buchthal analysis. Buchthal analysis was slightly modified before it was used as the gold standard for new multivariate diagnostical algorithms, based on principal component analysis and on MUP discriminant classification: muscle means of continuous variables were related to tolerance limits after adequate transformation. Chi-square tests were used for dichotomized variables, e.g., polyphasia. Sensitivity and specificity of the uni- and multivariate algorithms were compared for 539 muscles from patients with motor neuron diseases, neuropathies and myopathies and for 91 biceps brachii, rectus femoris and tibialis anterior control muscles. False positive results accumulated less than expected by repeat univariate tests for single MUP parameters, due to high correlation. Combination of single parameters to factor scores did not improve specificity. One advantage of factor analysis was that factor matrix and factor scores matched those of previous studies in spite of different input parameters, which may facilitate multicenter comparisons. Discriminant classification successfully separated neurogenic and myopathic conditions, even in myositic muscles and motor neuron diseases, where myopathic and neuropathic MUPs frequently intermingle. Discriminant classification may support expert decisions and add weight to EMG differential diagnosis.