Biostatistics · BEGINNER LESSON
Survival
In survival analysis, survival means the time from a starting point to an event such as death, relapse, recovery, or leaving the study. It is measured as time-to-event data, usually in days, weeks, months, or years.
What survival measures
In biostatistics, survival is not only about death. It is the measured time from a defined starting point, called the time of origin, to an event of interest. The event can be death, disease, relapse from remission, recovery, or another endpoint. The time must be a positive value and is usually recorded in hours, days, weeks, months, or years.
[1] Survival analysis: A primer for the clinician scientists - PMC [3] Biostatistics Series Module 9: Survival Analysis - PMC
How survival is analyzed
Survival analysis is the set of statistical methods used for time-to-event data. It describes patterns of event times, compares groups, and studies how explanatory variables relate to survival time. Two basic quantities are the survivor function S(t), which is the probability of surviving longer than time t, and the hazard function h(t), which is the instantaneous chance of the event occurring at time t among those who have survived up to that point.
[1] Survival analysis: A primer for the clinician scientists - PMC [3] Biostatistics Series Module 9: Survival Analysis - PMC
Why censoring matters
A major difficulty is that not every person has the event during the study period, and some people leave early. These incomplete records are called censored observations. Standard survival methods can handle censoring, but only when censoring is noninformative, meaning it does not itself signal a worse or better future outcome. If censoring is informative, the usual methods can give biased results.
[1] Survival analysis: A primer for the clinician scientists - PMC [3] Biostatistics Series Module 9: Survival Analysis - PMC
Common methods and their limits
Life table analysis and Kaplan–Meier analysis are common descriptive tools. Life tables group time into intervals, but the results can change depending on how the intervals are chosen. Kaplan–Meier re-estimates survival each time an event occurs and is widely used for survival curves, but it mainly handles one factor at a time and does not by itself perform multivariable analysis. The log-rank test compares survival curves without assuming a specific survival distribution.
[1] Survival analysis: A primer for the clinician scientists - PMC [3] Biostatistics Series Module 9: Survival Analysis - PMC
Mechanism in simple terms
Think of a study as starting a clock at a baseline date such as diagnosis or surgery. Each participant is followed until the event happens or until observation ends. The analysis uses the observed times and the censored times together to estimate how likely survival is over time and how quickly events occur among those still at risk.
[1] Survival analysis: A primer for the clinician scientists - PMC [3] Biostatistics Series Module 9: Survival Analysis - PMC
SEE THE IDEA
How survival analysis works
Start the clock at a defined origin such as enrollment or diagnosis.
Track each person forward in time until an event or study end.
Record either the event time or the time when complete information stops.
Use life table or Kaplan–Meier methods to estimate survival over time.
Baseline time leads to follow-up, which ends in either an event or censoring; the observed times feed survival estimates.
Survival analysis follows people from a starting point and uses both event times and censored observations to estimate time-to-event patterns.
TRY IT YOURSELF
Test your understanding
Choose an answer, then check the explanation.
PUT IT TOGETHER
Match each concept to its meaning
Drag a label onto its description, or select a label and then select a description. This also works with a keyboard.
4 pairs to match
Sources and further reading
This general lesson was generated from public source material. It explains a concept; it does not interpret a private paper or provide medical advice.