Diagnostic accuracy of heart rate variability in screening of obstructive sleep apnea

Purpose: to define the informative value of heart rate variability (HRV) when included in the logistic regression model with clinical dates to improve screening for OSA. Materials and methods. The study involved 124 people, who had positive test on an Epworth`s scale of daytime sleepiness. The cardio-respiratory monitoring and ECG monitoring were performed for all participants. 2 Groups were allocated as a result of cardio-respiratory monitoring: primary (77 people – the apnea-hypopnea index (AHI) > 5) and control (47 – AHI < 5). Both groups were similar by gender and age. Results. It has been found that the majority of HRV in the main group have a significant difference with the control group. There was an abnormal increase in the spectral parameters VLF, LF, LFN, both day and night. Power HF, SDNN, and RMSSD in the study group was significantly lower than the control group. The five variables were selected as the most informative by the method of gradual inclusion of each parameter in the logistic regression model: SDNNnight, BMI, gender, LFnight, RMSSDnight (the value of the area under the ROC – curve was 0.805). This model had sufficient sensitivity (84 %) and specificity (63 %) for the identification of OSAS. The scoring system was developed on the basis of the final regression model. According to the results, it has been found that the presence of 3 or more points on our scale indicates a need for cardio-respiratory monitoring. The sensitivity of the scoring system was 86 %, specificity – 69 % (AUC 0.841). Conclusions. The strongest connection with the AHI has been established for five variables: SDNNnight, BMI, gender, LFnight, RMSSDnight. The scoring system, which was built with the use of these variables, had sufficient sensitivity (86 %) and specificity (69 %) (AUC 0.841). The presence of 3 or more points according to the scoring system indicates a need for cardio-respiratory monitoring.