Suitability of time estimates for survival analysis of endodontic treatment

Abstract

Introduction: This study aimed to determine the most suitable estimates of the event date for Kaplan-Meier (K-M) survival analysis for the assessment of endodontic treatment outcomes. Survival probability computed using the Weibull accelerated failure time (WAFT) models based on interval-censored data was used for comparison. Methods: A sample of root canal-treated teeth was included for survival analysis using the K-M and WAFT models. Treatment outcomes were categorized for the presence or absence of periapical healing and tooth survival. For the K-M analysis, the following time estimates between a visit when failure was diagnosed and the preceding visit were tested: the one-quarter point (T1), the half-way point (T2), the three-quarter point (T3), the date of diagnosis (T4), and the geometric mean (T5). Results: The survival curve plotted from T2 consistently showed the least discrepancies in trend when compared against the WAFT curve. Those curves generated from T1 to T5 were significantly different from each other for periapical healing but not for tooth survival. Conclusions: T2 is the most reliable date for K-M analysis for measuring endodontic outcomes, especially when periapical healing is concerned. T4 would be a reasonable estimate for assessing tooth survival. Further research aiming at appropriate statistical exercise with simulations followed by proper goodness of fit tests should be considered. Copyright © 2013 American Association of Endodontists.