We agree with Asero and colleagues that cross-sectional studies such as ours cannot determine the temporality nor the natural history of the relationship between smoking (or reduced lung function) and systemic inflammation, an important limitation of the study, which we pointed out in the limitation section of our article.1However, cross-sectional studies can ascertain risk factors through measures of association.2 Thus, the statement that “reduced lung function is a risk factor” is fully justified by the methodology we used.1–2 We also agree that factors such as airway hyperresponsiveness (AHR), endothelial activation, and neutrophil recruitment may be involved in effecting systemic inflammation in smokers and nonsmokers with reduced lung function. However, these factors may not be confounders; rather, they may be part of the causal pathway(s) linking reduced lung function with systemic inflammation. For example, in the Lung Health Study,3which evaluated > 4,200 participants with mild COPD with serial methacholine challenge tests over 5 years, the investigators found that even among participants who stopped smoking, airway responsiveness increased during the follow-up period. This indicates that there are factors other than cigarette smoking that are responsible for AHR in individuals with reduced lung function. There are good experimental data to indicate that airway inflammation and remodeling are important determinants of AHR.4–5 Airway inflammation may also relate to systemic inflammation.6Thus, it would be misleading and erroneous to consider AHR as a confounder in the relationship between reduced lung function and systemic inflammation; it is likely to be part of its causal pathway. Finally, we disagree with the assertion of Asero et al that there was C-reactive protein (CRP) misclassification because we considered CRP levels > 2.1 mg/L to be elevated. In the general population, the geometric mean of CRP for individuals 55 to 64 years of age is between 1.6 and 2.2 mg/L.7By taking 2.1 mg/L as the cutoff threshold, we in effect used the median CRP value of the general population in dichotomizing the study sample, an approach that is widely accepted and commonly used to dichotomize continuous variables for analytic purposes in the medical literature.8 This approach strikes a reasonable balance between validity and efficiency. If we were to take the suggestion of Asero et al and use 10 mg/L as the cutoff value, we would significantly compromise the statistical power and the efficiency of the study without improving the validity of the findings.8 This approach therefore is best avoided.