RESUMEN
PURPOSE: Modern radiation oncology treatments are typically associated with a certain level of acute or long-term complications. Whenever a new treatment is introduced into clinical practice, a clinician may have reason to question whether certain side effects occur more frequently in his or her patients than reported in the published literature. However, the clinician, lacking rigorous statistical methods to answer this question, may not draw the correct conclusion until treating a larger number of patients than necessary. METHODS AND MATERIALS: We propose a Bayesian framework to calculate (1) the probability that the rate of an adverse event (AE) is above a certain threshold and (2) the number of AEs to be observed before we can conclude with a certain confidence level that the AE rate is above a threshold. Both questions will be answered based on the formulation of a posterior distribution of the AE rate given the observed data. The proposed method does not rely on large sample assumptions, and the computation of the solution is quick and straightforward. The proposed methods are illustrated in 2 clinical scenarios. RESULTS: We tabulated the probability that an AE rate is greater than a threshold for a given number of patients experiencing events from an observed number of patients. In addition, we developed a free online interactive tool to facilitate the implementation of the method. CONCLUSIONS: The proposed method, including the presented tables and the online tool, provides clinicians with a practical means of determining if his or her patients are experiencing a greater (or lower) number of complications than expected. How to use the statistical result to guide a clinical decision depends upon the specifics of the clinical condition and the available treatment alternatives.