Sensitivity and specificity of procedures for the differential diagnosis of epileptic and non-epileptic seizures: a systematic review.

OBJECTIVES: Non-epileptic seizures (NES) present a considerable challenge in clinical practice. This paper reviews published evidence for the reliability of a number of procedures for the differential diagnosis of NES and epilepsy. METHODS: Papers identified from MEDLINE and PsychInfo Databases (1980-2001) and additional hand searches were independently reviewed using methods for evaluating evidence in systematic reviews [Liddle, J., Williamson, M. Irwig, L. Method for evaluating research guideline evidence. New South Wales Department of Health; 1996 [State Health publication no. (CEB) 96-204]; SIGN. An introduction to SIGN methodology for the development of evidence based clinical guidelines. Scottish Intercollegiate Network; 1999]. Included studies had to have an NES group and a control group of people with epilepsy (each n> or =10), allocated using EEG linked video-recording of concurrent behaviour, and sensitivity and specificity values had to be stated or be calculable. RESULTS: Thirty-three papers were identified, of which 13 satisfied criteria. Excluded studies are briefly described. Those retained comprised a range of procedures [seizure induction, MMPI assessment, physiological assessment (prolactin, SPECT), pre-ictal pseudosleep, and ictal/post-ictal characteristics]. No procedure emerged with both high sensitivity and specificity and adequately replicated findings, although high levels of specificity were more commonly reported than high levels of sensitivity. This suggests that procedures were generally better at excluding a possible diagnosis. CONCLUSIONS: No procedure attains reliability equivalent to EEG video-telemetry. Further rigorous evaluation, using standardised and replicable methodologies, is required. The range of symptoms presented in NES suggests that a multi-method approach may be required. This too would require evaluation