Survival analysis is a statistical calculation assessing the duration of time that passes before death occurs. While survival is one of the most common endpoints in medical research and sounds straightforward — how long did someone live —it can actually be an incredibly complex concept. This is especially true when it comes to how survival is tracked in cancer research.
Deeper dive
Consider this selection of different survival endpoint types from the National Cancer Institute glossary:
- Cause-specific survival
- Disease-free survival
- Disease-specific survival rate
- Event-free survival
- Five-year survival rate
- Mean survival
- Median overall survival
- Median survival
- Relapse-free survival
- Relative survival rate
Other medical specialties, such as cardiology, may have more specialized types of survival that they track, so it’s important to pay attention in a medical study to precisely what kind of “survival” the researchers are measuring and what the potential limitations are of that endpoint.
Survival analysis is how survival is calculated: “a collection of statistical procedures for data analysis, for which the outcome variable of interest is time until an event occurs,” according to the excellent paper “Survival analysis in clinical trials: Basics and must know areas.” It’s impossible to go into all the nuances and details of survival analyses here, but there are some key questions to consider when reporting on a study that uses any survival metric as an endpoint:
When does the clock start? What specific events or conditions are they tracking (e.g., event-free survival vs. overall survival)? What confounders might interfere with survival and are they accounted for? Is the statistical method they are using to calculate and to compare survival statistics appropriate? (That one is a question for your biostatistician sources.) How are people lost to follow-up or who with withdraw from the study accounted for?
If in doubt, ask the researchers and your outside sources, including biostatisticians, to help you understand precisely how survival is being measured and used in the study.