2002 Abstracts

BACKGROUND: Dillard et al has published equations for predicting PaO₂ at altitude for patients with COPD and those with normal lung function. (Chest 1995;107:352). The purpose of this study was to validate the published results and estimate the error of predicting values for individual subjects.

METHOD: Outpatients with interstitial lung or pulmonary vascular disease and concerns about hypoxia during flight were studied. Steady state PaO₂ was measured at sea level and during a hypoxic inhalation test, breathing 15.1% oxygen (simulating PaO₂ at 8,000 ft). Gas was delivered with a Downs high flow generator on a nitrogen cylinder entraining room air. F_IO₂ was measured with a MaxO₂ oxygen analyzer calibrated with 5.1%, 10%, 15.1% and 19.8% precision blended gas. Dillard regression equations predicted PaO₂ at altitude using PaO₂ at sea level only, PaO₂ with FEV₁ % predicted or PaO₂ with FEV₁/FVC % predicted. Error intervals (Respir Care 1996;41:1090) were defined as mean difference (predicted PaO₂ ? actual PaO₂ at simulated altitude) plus or minus 2.659 standard deviations of the difference (based on Dillard?s sample size of 42).

Results: Nine patients were studied. Results were comparable with those published by Dillard et al. Data below give predicted minus measured in mm Hg:

Because the Dillard data were confirmed, we used their results (with a larger n) to construct error intervals for prediction error when estimating the PaO₂ at altitude for an individual patient. Intervals indicate range of values for 95% of estimates at 99% confidence level:

Conclusions: Our results were nearly identical to those of the Dillard study and confirmed that including FEV₁/FVC reduces prediction error. However, predictions for individual patients may be in error by as much as 10 mm Hg at best. Therefore, if the predicted PaO₂ is marginal, a hypoxic inhalation test may be indicated.