An observational study examining the relationship between oral contraceptive use and venous thromboembolism (VTE) in 2000 Indian women reports an unadjusted odds ratio of 3.2 (95% CI 2.1–4.8). After adjustment for age, smoking, and obesity, the odds ratio becomes 2.8 (95% CI 1.9–4.1). Which feature best distinguishes selection bias from confounding in this context?