|
Abstract:
|
An important question in the assessment of a clinical trial is to determine whether treatment cures (some) patients. On a similar note, in the public health field, it is important to assess the amount of disease burden that could be eliminated or reduced if risk factors were eliminated or reduced. Cure/immune models, used when long term survivors are present, consist of two parts; an accelerated failure time model for survival time of susceptible individuals and a logistic model for estimating the cure/immune proportion. Our simulations demonstrated that the true PAR is underestimated by a Cox PH model in the presence of cured/immune individuals. In addition, PAR estimates increase (or decrease) with the increasing (or decreasing) levels of the risk factor, but remain approximately unchanged (as expected) with increase or decrease in the cured/immune proportion.
|
