Modality-Specific Heart Rate Differences at Ventilatory Thresholds During Unified Ramp Testing

Overview

The purpose of this study was to investigate the heart rate bias between cycling and treadmill exercise at ventilatory thresholds. This was a randomized crossover trial involving 23 healthy participants. Each participant performed both cycling and treadmill exercise tests in a randomized order. The primary outcome was to determine the difference in heart rate response at the first and second ventilatory thresholds between the two exercise modalities. We hypothesized that there is a fixed bias in heart rate response, which may have implications for exercise prescription in clinical and rehabilitation settings.

Participants Twenty-three healthy adults (12 men, 11 women; age 30.8 ± 5.6 years; BMI 21.8 ± 2.6 kg/m²) were recruited via institutional advertisements at a university hospital. Inclusion criteria were: age ≥ 20 years, familiarity with cycling and treadmill exercise, and a negative Physical Activity Readiness Questionnaire response. Exclusion criteria were: history of cardiovascular, pulmonary, or metabolic disease; musculoskeletal injuries limiting exercise range of motion; or medications affecting heart rate or ventilation. Participants fasted for 3 hours, abstained from vigorous exercise, alcohol, and caffeine for 24 hours prior to testing, and provided written informed consent. Data collection was conducted between February 2024 and November 2024. The protocol was approved by the Institutional Review Board of National Taiwan University Hospital (IRB Registration: 202311036RIND). Experimental Design A randomized crossover design was employed. Participants completed two symptom-limited CPET sessions on a cycle ergometer (VIAsprint, Ergoline, Germany) and motorized treadmill (Ergosprint, Ergoline, Germany), separated by 7-14 days and conducted at the same time of day (±2 hours) to control for circadian effects. The randomized order minimized potential order effects and ensured that fatigue from the first test did not systematically bias results. Cardiopulmonary Exercise Testing Protocol Gas Exchange Measurement and Safety Monitoring Breath-by-breath gas exchange was measured using a metabolic cart (Vmax Encore, CareFusion, USA), calibrated before each test with standard gases and a 3-L calibration syringe. Heart rate was monitored continuously via 12-lead electrocardiography (Cardiosoft, GE Healthcare, USA), providing greater temporal resolution than telemetric methods. Blood pressure was measured at rest and every 2 minutes during exercise to monitor safety and detect cardiovascular abnormalities. Unified Ramp Protocol The central innovation was implementing a unified ramp protocol on both modalities to eliminate kinetic confounding. Cycle Ergometer: Following a 3-minute unloaded warm-up (0 W), continuous ramp increases began at 20 W·min-¹ (men) or 15 W·min-¹ (women), targeting an 8-12 minute test duration to optimize threshold identification within a physiologically relevant window. Treadmill: Using the Porszász algorithm, simultaneous adjustments to speed and grade produced metabolically equivalent work rate increases (20 or 15 W·min-¹), matched to cycling. Warm-up consisted of 3 minutes at 3-4 km/h, 0% grade. The Porszász protocol mathematically calculates the non-linear combination of speed and grade increases needed to produce a linear rise in metabolic demand, ensuring identical oxygen uptake kinetic demands despite different biomechanical constraints. Maximal Effort Verification Participants exercised to volitional exhaustion or until safety criteria were met. Maximal effort was confirmed by meeting ≥2 of the following criteria: respiratory exchange ratio (RER) ≥ 1.10; heart rate within 10 bpm of age-predicted maximum (220 - age); rating of perceived exertion (RPE) ≥ 17 on the Borg 6-20 scale; or VO₂ plateau (≤150 mL·min-¹ increase despite increasing workload). Threshold Determination Breath-by-breath gas exchange data were binned into 5-second moving averages to attenuate noise while preserving dynamic resolution at the threshold transition point. VT1 and VT2 were identified independently by two blinded, experienced raters using consensus criteria (V-slope method, ventilatory equivalents, end-tidal gas tensions). Blinding was maintained by removing all identifiers before threshold determination. Disagreements (2 cases) were adjudicated by a third blinded expert. Inter-rater reliability was excellent (ICC = 0.94 for VT1; ICC = 0.96 for VT2). Variables were extracted as 30-second averages centered on the identified threshold time point to ensure stability. Statistical Analysis Primary and Secondary Outcomes Primary outcome: heart rate difference at VT1 and VT2 between modalities. Secondary outcomes: absolute and relative VO₂, EqO₂, EqCO₂, and O₂ pulse. Data Analysis Methods Data normality was assessed using the Shapiro-Wilk test. Paired t-tests compared variables between modalities, with two-tailed significance testing at α = 0.05. Intraclass correlation coefficients (ICC; two-way mixed-effects, absolute agreement) assessed measurement reliability. Measurement error was quantified using the coefficient of variation (CV = SD of differences/mean), which normalizes variability to enable comparison across variables with different scales. Bland-Altman Regression Analysis for Proportional Bias Bland-Altman analysis with linear regression tested for fixed versus proportional bias-the core statistical innovation distinguishing this study. The difference between modalities (Cycling - Treadmill) was regressed against the mean of the two measurements. A non-significant regression slope (p \> 0.05) indicated fixed bias, whereas a significant slope (p \< 0.05) indicated proportional bias. This approach directly tests whether correction factors should be constant across participants or adjusted based on individual physiological capacity. All statistical analyses were performed using SPSS Statistics Version 20 (IBM Corp., Armonk, NY) for ICC and t-tests, and GraphPad Prism Version 10 (GraphPad Software, San Diego, CA) for Bland-Altman plots.