Quantifying bias in epidemiologic studies evaluating the association between acetaminophen use and cancer