Background  Dysglycemia leads to autonomic impairment in patients, and heart rate variability (HRV) has been widely used as a non-invasive method for evaluating autonomic nervous function. However, most studies focus on the linear parameters of HRV, so further studies are needed to investigate the effect of dysglycemia on non-linear parameters of HRV.

  Objective  To investigate the effect of dysglycemia degree on the changes in the time domain, frequency domain and non-linear parameters of HRV.

  Methods  Totally 204 patients who were hospitalized in the Hyperbaric Oxygen Ward of the First Medical Center, Chinese PLA General Hospital from March 2018 to January 2020 were included in this study. The normal glucose level (NGL) group included 85 cases aged (56.9 ± 11.4) years with 49 males, impaired glucose tolerance (IGT) group included 32 cases aged (57.0 ± 10.1) years with 25 males, type 2 diabetes mellitus (T2DM) group included 87 cases aged (59.6 ± 10.0) years with 48 males. Continuous physiological data were collected using a follow-up physiological monitoring system (SensEcho). Parameters of 24h admission HRV such as time domain, frequency domain, Poincaré plot, Heart Rate Asymmetry (HRA) and Multiscale Entropy (MSE) at 4h during the day were calculated for the three groups and compared using variance analysis.

  Results  The standard deviation of the normal-to-normal RR interval (SDNN), the coefficient of variation (CVNN) and the triangular index (HTI) of the HRV linear parameters decreased progressively in the three groups (P＜0.001), and the ultra-low frequency (ULF) and very low frequency (VLF) were lower in the IGT group compared to the NGL group (OR95% CI: 0.995 0.992, 0.998, P=0.001 for ULF; and OR(95% CI: 1.002 1.001, 1.004, P=0.003 for VLF). For the non-linear parameters, the overall contribution variance (SDNNa) of heart rate acceleration in HRA (OR95% CI: 0.954 0.933, 0.975, P＜0.001) and overall complexity index (Area1_20) of MSE (OR95% CI: 0.857 0.775, 0.947, P＜0.01) were lower in the T2DM group compared to NGL group; and the high frequency power (HF) (OR95% CI: 0.994 0.991, 0.997, P＜0.001) and low frequency to high frequency ratio (LF/HF) (OR95% CI: 0.442 0.292, 0.670, P＜0.001) were also significantly lower in the T2DM group compared to the IGT group; ULF was negatively correlated with blood glucose level, total glycosylated hemoglobin and 2h glucose tolerance level (r=-0.302, P＜0.001; r=-0.254, P＜0.05; r=-0.252, P＜0.05).

  Conclusion  Linear parameters such as SDNN, CVNN and HTI in the time domain decrease with increasing severity of impaired glucose metabolism, and the decrease in ULF and VLF in the frequency domain may have started in pre-diabetes, while LF reduces significantly only in diabetes, ULF is negatively correlated with the three glucose metabolic indices, and LF/HF shows larger values in pre-diabetes. The non-linear SDNNa and Area1_20 are significantly lower in T2DM patients. It is suggested that dysglycemia may lead to a predominance of sympathetic activity in patients with IGT, while an overall reduction in autonomic activity and heart rate complexity, and an acceleration in patients with T2DM, which in turn affects the cardiovascular regulatory mechanisms of patients. These HRV parameters may be useful indicators for the staging of dysglycemia and the assessment of outcome.