A public health researcher in Delhi conducts a study to investigate the association between coffee consumption and myocardial infarction (MI) in adults aged 40–65 years. The study recruits 500 cases (patients with MI) from a tertiary hospital and 500 controls (age-matched, hospital visitors without MI). Cases report significantly higher coffee consumption (mean 4 cups/day) compared to controls (mean 1.5 cups/day), with an odds ratio of 2.8 (95% CI: 2.1–3.7). However, when the researchers stratify the analysis by smoking status, the association disappears in both smokers and non-smokers separately. What type of bias or confounding is most likely responsible for the observed association?