Correlated Poincare indices for measuring heart rate variability

Abstract

Poincare indices are usually applied to HRV to summarise long data sets collected over 24 hrs. Many applications of HRV are interested in dynamic, short term changes (<1 min). This study uses Poincare indices published through the 1990's to the present, to determine which of them are correlated over the short term (25 beats). Dynamic changes were observed in 12 subjects preoperatively receiving fentanyl and midazolam sedation with ECG collected for 5 mins before and 5 mins after fentanyl administration. Poincare indices with a strong correlation (r>0.85) between the indices for each of the 12 subjects (p<0.001) (particularly with the common measures SDNN, RMSSD, pNN50 and meanRR) were identified. These indices will not be used for further investigation of dynamic effects of fentanyl and midazolam, two sedative drugs used in anaesthesia and intensive care. Indices that proved less suitable for short term analysis (eg, presence of outliers, inability to produce a valid index with smaller number of beats) were also identified. A shortlist of Poincare indices that do not correlate strongly with commonly used measures may prove interesting in determining dynamic characteristics of the effect of sedative drugs on autonomic nervous system activity.