Arbitrary thresholds of natural variability and severity should not be used to diagnose chronic obstructive pulmonary disease