Nonrandom variability of respiration during sleep in healthy humans

Breath-to-breath variability is not purely random but is, instead, characterized by correlations on short- and long-term scales. Short-term correlations might reflect intact metabolic-control mechanisms. To investigate whether the higher variability of breathing during rapid eye movement (REM) compared to non-REM (NREM) sleep is of random or nonrandom nature--reflecting an altered respiratory control--short-term and long-term correlations of respiratory drive and timing were determined.

A full-night polysomnogram with a pneumotachograph attached to a full-face mask was performed. For each breath during NREM and REM sleep, respiratory components were analyzed based on the quantitative airflow.

Data collection took place in the sleep laboratory.

Twenty-nine healthy subjects (age, 25.8 +/- 3.1 years).

Long-term correlations are practically absent in respiratory timing and drive components during NREM sleep, whereas they are present during REM sleep. Short-term correlations are present in respiratory drive, tidal volume, and minute ventilation during both NREM and REM sleep. In all timing components, additional short-term correlations are absent.

We conclude that from NREM to REM sleep, short-term regulation of respiratory drive remains strongly metabolically controlled and clearly different from the short-term regulation of the rhythm-generating function. Regulation of respiratory timing and drive during REM sleep is characterized by additional long-term correlations. We speculate that this is the result of cortical influences during phasic REM sleep. Thus, the variability of breathing during REM sleep contains a nonrandom component, such that breathing components remain dependent upon each other even with large time lags between components.