A 52-year-old male factory worker from Mumbai presents with a 6-month history of persistent cough and dyspnea. He has worked in a textile mill for 28 years without protective equipment. Chest X-ray shows bilateral upper lobe fibrosis. A study is initiated to examine the association between textile dust exposure and pulmonary fibrosis in 500 workers from the same factory. The researchers find a strong positive association (RR = 4.2, 95% CI 2.8–6.1). However, upon further analysis, they discover that workers with longer tenure (>25 years) were more likely to have been exposed to higher dust concentrations in the past, and these same workers were also more likely to be current smokers. When smoking status is controlled for in the analysis, the RR drops to 1.8 (95% CI 0.9–3.5). Which of the following best describes the relationship between smoking and the observed association between dust exposure and pulmonary fibrosis?

Explanation

## Identification of Confounding **Key Point:** A confounder is a variable that is associated with both the exposure and the outcome, and distorts the true causal relationship. Confounding is identified when adjustment for a third variable substantially changes the measure of association. ### Analysis of the Scenario In this study: - **Exposure:** Textile dust exposure - **Outcome:** Pulmonary fibrosis - **Suspected confounder:** Smoking status **Associations present:** 1. Smoking is associated with dust exposure (workers with longer tenure → more past dust exposure AND more likely to be smokers) 2. Smoking is independently associated with pulmonary fibrosis (well-established causal relationship) 3. These associations are independent of the dust-fibrosis pathway ### Evidence of Confounding When smoking is controlled for: - RR changes from 4.2 to 1.8 (>10% change = clinically significant confounding) - The 95% CI shifts from 2.8–6.1 to 0.9–3.5 (now includes 1.0, suggesting the association may not be causal after adjustment) - This substantial change indicates smoking was inflating the observed association **High-Yield:** The magnitude of change in the effect estimate (RR dropped by >50%) is the hallmark of confounding. Confounding biases the association away from the null, making weak associations appear strong. ### Why NOT the Other Options **Effect modification** would show different RRs in different strata (e.g., RR in smokers ≠ RR in non-smokers), but the question doesn't describe stratified results. **Mediation** would mean smoking lies in the causal pathway (dust → smoking → fibrosis), but smoking precedes dust exposure chronologically in this cohort, making it impossible to be a mediator. **Intermediate variables** (mediators) should NOT be controlled for; confounders SHOULD be controlled for. Smoking is clearly a confounder here. ## Mnemonic for Confounding vs. Mediation **"BEFORE the exposure = Confounder; AFTER the exposure = Mediator"** - Smoking status existed before/independent of dust exposure → Confounder - If dust caused smoking, then smoking caused fibrosis → Mediator (not applicable here) **Clinical Pearl:** In occupational epidemiology, smoking is one of the most common confounders when studying respiratory outcomes. Always assess smoking status as a potential confounder in lung disease studies.