Regency EF
Interval-Censored Survival Analysis: A Practical and Underused Tool for Observational Data (304088)
*Travis Snyder, Imgen, SimonMed, Touro University Nevada*Cheryl Vanier, Touro University Nevada
Keywords: interval censoring, survival analysis, medical data
Survival analyses typically assume that the time of an event (TE) is precisely known. However, interval censored (IC) data, when TE is localized to a time interval, are common, and data which could most appropriately be analyzed using IC survival analysis techniques are often analyzed in a survival analysis that does not consider the censoring. We will demonstrate the use of IC survival analysis applied to a retrospective radiology data set, followed by a simplified and clear presentation of results in comparison with results without IC. The scientific purpose of the study was to provide physicians with prognostic information regarding longevity of particular symptoms up to four years after a traumatic brain injury (TBI), based on whether patients experienced loss of consciousness at time of injury. Analyses which did not incorporate IC produced misleading positive results which could not be supported. IC survival analysis is an important statistical tool to help summarize retrospective data and formulate hypotheses for further study. This technique is relevant to research in the fields of engineering, basic sciences, psychology, etc., and can be implemented in R, SAS, and Stata.