Heart rate variability (HRV) considered as an indicator of the autonomic activity, shows differences between NREM and REM sleep. However, when evaluating during NREM sleep, some authors evaluate without considering the different stages, while others do it. There are discrepancies between values obtained from stages N2 and N3. Our objective was to evaluate whether differences exist between stages N2 and N3 based on several metrics of HRV that reflect the autonomic activity. Additionally, we analyzed the structure of these metrics through principal component analysis (PCA) at each sleep stage. Polysomnograms were recorded in 24 healthy subjects. We found significant differences in the LF and THM metrics with higher values in stage N2. The relative power of HF metric showed a significant difference with higher value in stage N3, however, this metric showed inconsistencies related to other HF representations. Two factors were extracted with PCA and their structure was stable across sleep stages. One factor accounted for 92% of the variance during stage N2 with moderate to high loadings for metrics most strongly associated with the parasympathetic activity. A second factor accounted for 91% of the variance during stage N3 with moderate to high loadings for metrics associated with parasympathetic and blood pressure control. Physiological mechanisms regulated by central commands that may involve neurons in the hypothalamus, pons, and midbrain may explain the differences in LF and THM metrics. The combination of stages N2 and N3 as NREM should be done according to the type of HRV metrics used.
