Actuarial and Kaplan-Meier survival analysis: There is a difference

While attending the Seventy-ninth Annual Meeting of The American Association for Thoracic Surgery in New Orleans, I was dismayed to see the almost universal misuse of the term “actuarial survival.” There are two general types of analysis for survival information: actuarial and Kaplan-Meier. An actuarial analysis should be performed when the actual date of a survival event is unknown. The known information is that the event occurred between time tn and time tn+1. Actuarial analysis is carried out at specific time intervals (6 months, 1 year), and the resulting graph will step only at those intervals. As the actual failure time is only approximated by the end point, the convention is to attribute a survival time of the fully completed intervals plus half the time of the interval during which the event occurred. Examples of events needing actuarial analysis include population-based death rates and disease-free survival. Kaplan-Meier analysis is used when the actual date of the end point is known. End points not reached are treated as censored at the date of last follow-up for the analysis. Kaplan-Meier analysis is undertaken at each survival event, death, or censoring, and the graphs will step at each failure time and may or may not be drawn to show the location of censored observations. Examples of an appropriate event for Kaplan-Meier analysis would be postoperative survival when the date of deaths is known. Most of the misuse of the term “actuarial survival” came during presentation of data with well-known end points and graphs that clearly reflected a Kaplan-Meier analysis. To mislabel the precise survival estimates with the term “actuarial survival” suggests a limited understanding of survival analysis. 12/8/102480