Explained variance for blood gases in a population with COPD.

INTRODUCTION: Variation of blood gas levels in chronic obstructive pulmonary disease (COPD) patients has not been extensively reported and there is limited knowledge about predictors of chronic respiratory failure in COPD patients.
OBJECTIVES: The aim of this study was to identify predictors of hypoxemia, hypercapnia and increased alveolar-arterial oxygen difference in COPD patients. We hypothesized that prediction of arterial blood gases will be improved in multivariate models including measurements of lung function, anthropometry and systemic inflammation.
METHODS: A cross-sectional sample of 382 Norwegian COPD patients, age 40-76, Global Initiative for Chronic Obstructive Lung Disease stage II-IV, with a smoking history of at least 10 pack-years, underwent extensive measurements, including medical examination, arterial blood gases, systemic inflammatory markers, spirometry, plethysmography, respiratory impedance and bioelectrical impedance. Possible predictors of arterial oxygen (PaO(2)), arterial carbon dioxide (PaCO(2)) and alveolar-arterial oxygen difference (AaO(2)) were analyzed with both bivariate and multiple regression methods.
RESULTS: We found that various lung function measurements were significantly associated with PaO2, PaCO(2) and AaO(2). In addition, heart rate and Fat Mass Index were predictors of PaO(2) and AaO(2), while heart failure and current smoking status were associated with PaCO(2). The explained variance (R(2)) in the final multivariate regression models was 0.14-0.20.
CONCLUSIONS: With a wide assortment of possible clinical predictors, we could explain 14-20% of the variation in blood gas measurements in COPD patients.