Abstract
In normal subjects, adjusted to a circadian environment, half-hourly plasma cortisol levels over a 24-h span (time series, TS) outline a circadian sinusoidal pattern that may be masked by superimposed short-lived oscillations. The periodogram objectively defines the features (amplitude, phase) of this pattern in individual TS and, more efficiently, in the average and concatenation of homogeneous TS. Concatenation is a procedure by which homogeneous short TS are attached one after the other. Homogeneity is inferred by variance analysis and by a new measure, h, with statistical significance, derived from the spectrum of the concatenated TS. Through concatenation, autocovariance for periodicity identification becomes possible. The TS from three groups of volunteers are highly homogeneous and have a predominant, statistically significant amplitude with peak time near 1100 h. The patterns from patients with Cushing's disease differ markedly in their means and periodic features; by the above criteria, we identify a subgroup with significant h measure and 24-h amplitude. This analysis provides a tool for investigation of the circadian clock in man.

This publication has 2 references indexed in Scilit: