Percentile-based versus conventional thresholds in cardiopulmonary exercise testing: a retrospective cross-sectional study.

INTRODUCTION

Determining whether a value falls within the expected range is a central step in cardiopulmonary exercise testing (CPET) interpretation. Conventional thresholds based on absolute or percentage-predicted values are widely used but fail to capture population variability. This study evaluated the validity of percentile-based thresholds derived from the Study of Health in Pomerania (SHIP) for a Western population and compared their impact on classification with conventional thresholds in individuals without known cardiorespiratory disease and patients with cardiorespiratory disease.

METHODS

We retrospectively analysed 886 maximal CPETs from 294 individuals without known cardiorespiratory disease (control group) and 592 patients with COPD, interstitial lung disease, heart failure or pulmonary hypertension. Percentile-based thresholds were evaluated for suitability by comparing abnormality rates with the expected 5% in the control group for peak oxygen uptake (V'O2 peak), ventilatory equivalent for carbon dioxide (V'E/V'CO2 nadir values), peak heart rate (HRpeak), V'O2 /work rate (WR) slope and peak V'O2 /HR. Agreement, discordance, and cumulative impact of percentile-based versus conventional thresholds were assessed.

RESULTS

Percentile-based thresholds classified 4-8% of CPETs as abnormal in the control group, consistent with the expected 5%. In patients, conventional thresholds classified up to 14% more V'O2 peak values and 31% more HRpeak values as abnormal compared with percentile-based thresholds, whereas percentile-based thresholds identified more abnormal V'E/V'CO2 nadir values. Full agreement across all variables was observed in 82% of CPETs in the control group versus 46-58% in patients.

CONCLUSION

The SHIP-derived percentile-based thresholds are valid for Western populations, and their application leads to substantial differences in abnormality classification of CPET results compared to conventional thresholds, especially in patients.