Testing concomitancy between two physiological pulse series.

Abstract

Statistical methods for detecting synchronization of onsets of two physiological events, such as concomitant releases of prolactin (PRL) and luteinizing hormone (LH), are studied in this paper. Usually, the two events are measured regularly and simultaneously and the onset times are then determined by the changes in the values of the measurements. Owing to difficulties in determining the exact onset times, a leeway is allowed for counting the onset synchronizations in most physiological studies. Using such a relaxed definition of synchronization and needing to allow for a recovery time between two episodes make the traditional hypergeometric test of concomitance inappropriate. Based on a truncated geometric inter-arrival distribution, Clifton et al. have used simulation to construct a table of critical values for performing tests of significance of the observed number of coincidences in a series of 73 measurements. In this paper, the conditions under which their method can be used are examined from a statistical point of view. Methods for testing cases not covered by Clifton et al. are also described.