Abstract
Objective: Some patients with psychogenic nonepileptic seizures (PNES) remit when given the diagnosis. It is not realistically possible to test this potential therapeutic effect in an Randomized Controlled Trial (RCP) so we aim to statistically demonstrate it using the temporal relationship between the communication of the diagnosis and the timing of remission.
Method: Re-analysis of data from a study of PNES, where diagnosis was communicated, and outcomes recorded in 54 patients. Making conservative assumptions and using the binomial distribution, the Poisson distribution and the chi-squared test distribution, we calculated likelihoods of the null hypothesis: that communication of the diagnosis and remission of seizures had occurred in random temporal relationship.
Results: Remission occurred in the week following communication of the diagnosis in 15 out of 54 patients. The chi(2) test assigned this result a p value of <0.00001. Binomial and Poisson distribution cakulations also indicated that remission was highly unlikely to have occurred by chance and that, in a dataset similar to ours, was unlikely to be due to chance if occurring in more than 9 patients (16.7%).
Conclusions: We showed that the observed remissions were highly unlikely to be due to chance. Where an intervention is 'short and sharp' and the outcome can be measured with reasonable temporal acuity, then this type of method may provide an alternative to RCP methodology when the latter is impracticable. (C) 2019 Elsevier Inc. All rights reserved.