Reading the research · Annotated reading guide

How to read a canine survival curve

Read the axes, the risk table and the disappearing observations before judging a gap.

Start with the event

“Survival” analysis follows time until a defined event. Confirm what that event is. A graph of time until a diagnosis is not the same as a graph of time until death. Find where time starts: birth, enrollment, diagnosis or treatment can produce different interpretations.

Follow the steps

In a Kaplan-Meier plot, an event produces a downward step. A flat segment means no recorded event changes the estimate during that interval; it does not prove that no animal left observation. Censoring indicates that follow-up ends without the event being observed at that point. The method depends on assumptions about that incomplete information.

Use a fictional reading example

Suppose two invented groups begin with 100 dogs each. Near the far right of the graph, only five and six remain under observation and at risk. A visible gap there rests on much less information than a gap earlier in follow-up. These numbers illustrate how to read the display; they are not findings from a canine study.

Check the risk table and uncertainty

Look for the number still at risk below the graph and confidence intervals around the estimate. Be cautious about making a strong claim from a sparse tail. Also inspect the vertical scale: a cropped axis can make a small difference look dramatic.

Avoid a causal leap

A gap between observational groups does not prove the group label caused the difference. Check how groups were defined, which differences were considered and whether assignment was randomized. A curve is a summary of a study, not an individual forecast. Do not read off how long your dog will live from a population line.

Sources and further reading