During a cross-sectional survey on the association between smoking and chronic obstructive pulmonary disease (COPD) in a factory workforce, the investigator finds a weak association (PR = 1.3, 95% CI 0.95–1.78). The investigator suspects that occupational dust exposure may be confounding the relationship. The prevalence of dust exposure is 60% among smokers and 20% among non-smokers. What is the most appropriate next step to evaluate and control for this confounding?