Can RR intervals editing and selection techniques interfere with the analysis of heart rate variability?

• Main Author: Ribeiro G.D.S.
• Publication Date: 2018
• Other Authors: Neves V.R., Deresz L.F., Melo R.D., Dal Lago P., Karsten M.*
• Format: Article
• Language: eng
• Source: Repositório Institucional da Udesc
• dARK ID: ark:/33523/001300000q8f9
• Download full: https://repositorio.udesc.br/handle/UDESC/6154

© 2018 Associação Brasileira de Pesquisa e Pós-Graduação em FisioterapiaBackground: Oscillation between successive sinus beats or RR intervals, termed heart rate variability, is an important marker of autonomic function of the heart. However, its analysis may be influenced by the database recorded based on the occurrence of interference. Objective: To evaluate if the techniques of identification and editing of artifacts, as well as the selection methods of RR intervals, can interfere with heart rate variability analysis. Methods: The RR intervals of 56 subjects (30 aortic stenosis patients, 14 physically active individuals, 12 amateur athletes) were recorded for 10 min using a heart rate monitor. Values with differences greater than 20%, higher than three standard deviations or outside of the normal curve (95% confidence interval) were considered artifacts. These points were corrected through data replacement, adjacent, linear and polynomial interpolation, or excluded. Then, the 256 highest stability points and the last 5 min of recordings were chosen. The software programs, Kubios HRV and GraphPAD, were used to calculate and to analyze the indices of heart rate variability, respectively. Results: Strong agreement was observed among the identification algorithms; there was no difference between the correction techniques (p = 0.95); and the selection methods exhibited different sections (p < 0.01) with a direct influence on approximated entropy (p < 0.05). Conclusion: With short-term recordings, selection methods may interfere with the non-linear heart rate variability analysis. The confidence interval, the replacement by the average of previous data and the selection of 256 of the highest stability points of the signal seem to be the most adequate procedures to treat the data with prior to analysis.