Comparison of different baseline parameters for the prediction of an arterial oxygen tension (Pa,O2) at altitude of <6.6 kPa at rest using receiver-operating characteristic curves#
| AUC | 95% CI | Cut-off value | False-positive rate % | p-value | |
| MEF75 % pred | 0.91 | 0.79–1.01 | <31 | 9 | 0.01 |
| FEV1 % pred | 0.89 | 0.77–1.01 | <48 | 9 | 0.01 |
| Tiffeneau index¶ | 0.89 | 0.76–1.10 | <56 | 9 | 0.01 |
| MEF25 % pred | 0.84 | 0.70–0.98 | <13.5 | 17 | 0.004 |
| Pa,O2 baseline kPa | 0.86 | 0.71–1.00 | <9.5 | 23 | 0.004 |
| Sa,O2 baseline % | 0.85 | 0.71–1.00 | <94 | 28 | 0.03 |
| Pa,O2 after HIT kPa | 0.80 | 0.58–1.00 | <6.8 | 33 | 0.02 |
The cut-off values and false-positive rates correspond to a sensitivity of ≥75%. AUC: area under curve; CI: confidence interval; MEF75: maximal expiratory flow when 75% of the forced vital capacity (FVC) remains to be exhaled; FEV1: forced expiratory volume in one second; MEF25: maximal expiratory flow when 25% of the FVC remains to be exhaled; Sa,O2: arterial oxygen saturation; HIT: hypoxia inhalation test; % pred: percentage of the predicted value. #: the greater the AUC the better the predictive value of the baseline parameter (see fig. 1⇑); for example, in the present population, six of 36 (sensitivity 84%) patients with an FEV1 of <48% pred showed resting Pa,O2 at altitude of <6.6 kPa, whereas three of 36 patients (8.6%) showed Pa,O2 of >6.6 kPa; ¶; FEV1/FVC.