Patient-specific predictions for clinical decision-making

There are many ways in which clinicians make decisions regarding therapy for an individual patient. Invariably, the decision is supplemented by collateral information which is obtained from clinical trials. Prospective and retrospective clinical studies take an unnecessary adversarial position but each method has its place. Parametric survival analysis can produce patient-specific predictions of an event of interest for an individual patient based on the risk factor profile of that patient, and these methods are just as applicable to prospective clinical studies as they are to retrospective clinical studies. Parametric survival analysis can estimate the distribution of time to an event, estimate the phases of hazard (instantaneous risk of the event), determine risk factors associated with the event in each of the identified phases of risk, predict the time-related probability of an event for a specific patient based on his or her risk factor profile, and by risk adjustment compare the effectiveness of different therapies.