Objective: To form median somatosensory evoked potential (SEP) monitoring recommendations based on signal-to-noise ratio (SNR).
Methods: Two 1024-sweep right median SEP trials were recorded in 35 patients undergoing spine surgery. The SNR (signal power/noise power) and sweeps to reproducibility (<30% and <20% signal variation) were compared between the following derivations: cubital fossa (CF), Erb's point (EPi-EPc, EPi-M, EPi-Fz), cervical (C5S-EPc, C5S-AC, C5S-M, C5S-Fz), subcortical (CPi-EPc, CPi-M), and cortical (CPc-EPc, CPc-M, CPc-FPz, CPc-Fz, CPc-CPi, CPc-CPz), where M was the mastoid.
Results: Higher SNR produced markedly faster reproducibility. The CF derivation had very high SNR and single-sweep reproducibility. Of cortical derivations, CPc-CPz had highest mean SNR and fastest overall reproducibility (median 50 and 120 sweeps to <30% and <20% signal variation); occasionally CPc-Fz or CPc-CPi was better. Of Erb's point and cervical derivations, EPi-M and C5S-M had highest mean SNR and fastest reproducibility. Subcortical derivations had very low mean SNR and slow or non-reproducibility. High voltage EEG degraded cortical and subcortical derivation SNR and reproducibility in young children.
Conclusions: The highest SNR derivations should be used to speed surgical feedback; slower low-SNR derivations should be omitted. Consequently, the CF is the best technical control and CPc-CPz should be the standard cortical derivation, with CPc-Fz and CPc-CPi as alternates. EPi-M and C5S-M are the best Erb's point and cervical derivations, but are optional. Subcortical derivations should be omitted. A presence/absence criterion or SEP omission may be indicated for some young children.
Significance: The results should influence future guidelines.