Abstract

Purpose
The ventilatory threshold (VT) is a physiological turning point that can be used to guide for exercise prescription, as a tool to monitor response to an intervention and as a prognostic marker, but the presence of respiratory disease may limit the reliability of its measurement. This project aimed to determine the reliability of the assessment of the ventilatory threshold among human and computerized observers, in patients with chronic obstructive pulmonary disease (COPD) and controls.

Methods
VT was identified from incremental exercise testing graphs of 115 subjects (23 controls and 23 in each COPD severity class) by two human observers and a computer analysis, using the V-slope method and the VEM. Agreement between observers for VO2 at VT (VO2VT) and heart rate at VT (HRVT) were evaluated using intra-class correlation (ICC) for humans and Passing-Bablok regression analysis (human vs computer).

Results
For humans, ICCs for VO2VT were higher in controls [0.98 (0.97-0.99) both with V-slope and with VEM] than in COPD patients [0.72 (0.60-0.81) with V-slope and 0.64 (0.50-0.74) with VEM]. Human and computerized values of VO2VT were interchangeable in controls, but not in COPD patients. FEV1 and peak-ventilation were independent predictors of a lesser reliability of VO2VT. Inter-observer differences in HRVT ranged from 2±1 beats/minute (controls) to 10±3 beats/minute (GOLD 4).

Conclusions
In COPD, the reliability of human estimation of VO2VT is less in than in controls and not interchangeable with a computerized analysis. This should be taken into account when using VT in the clinical and research settings.