If the cause of death was falsely attributed to prostate cancer in those patients with prostate cancer who smoked, and since smokers are more likely to die, then more thorough ascertainment of prostate cancer in smokers would have occurred. Screening for prostate cancer at baseline may be less common in smokers than non-smokers, leading to elevated mortality from prostate cancer due to a delay in treatment. The Health Professionals cohort study did analyses which found that these two biases were unlikely to account for their results (Giovannucci et al 1999).