The important role of time series stationarity for agreement of ultra-short-term heart rate variability in ski mountaineers: a case series

Abstract Study aim: The validity of ultra-short-term (≤1 min) heart rate variability (HRV) analysis in sports science remains debated. Previous studies have not established criteria for acceptable agreement or considered the role of signal stationarity. The aim of this study was to evaluate the agreement between HRV calculated based on RR intervals (RRi) from 5-minute (criterion) and shorter durations in elite athletes, with emphasis on the impact of signal stationarity. Material and methods: Eight ski mountaineers underwent ECG recordings in a supine position. Criterion recordings were segmented into the first: 4-min, 3-min, 2-min, and 1-min intervals. HRV parameters (HR, RMSSD, RMSSD/mRR, lnRMSSD) were analyzed. Signal stationarity was assessed using the augmented Dickey-Fuller test. Agreement was evaluated using Bland-Altman analysis with the smallest worthwhile change defining acceptable differences and intraclass correlation coefficients. Results: HR and natural logarithm of RMSSD (lnRMSSD) calculated from 4-min recordings showed clinically acceptable agreement with criterion. Three athletes exhibited non-stationary RRi in 1-min and 2-min recordings. These non-stationary recordings demonstrated higher bias in HRV parameters compared to stationary recordings. Heart rate microstructure (acceleration/deceleration patterns) remained unchanged across different recordings. Signal non-stationarity in shorter recordings may affect agreement between HRV parameters in athletes. Conclusions: Ultra-short recordings (1-min) may not be reliable for all athletes or all HRV parameters. Individual variation in signal stationarity appears to be an important consideration. Practitioners should verify signal stationarity and define acceptable difference thresholds before adopting shortened recording protocols for HRV analysis.