Key takeaways:
The Kaplan-Meier curve is a statistical tool used to determine the probability of an event, like a cardiovascular issue or cancer progression, not occurring over a period of time.
By comparing the Kaplan-Meier curves of a treatment and a control group, researchers can determine if treatment reduces the probability of an unwanted outcome happening in the group.
When talking to your patients, it’s important for them to understand that Kaplan-Meier curves measure probability for a group — an individual’s response could be very different.
Statistics is a tough subject, and interpreting statistical tests used in clinical trials can be confusing for even the most seasoned healthcare professionals.
Even so, statistics forms the backbone of evidence-based medicine. Because of this, having a good understanding of the most common statistical tests can help us provide the best care to our patients.
We previously looked at p-values, and here we’ll look at another commonly used statistical tool: Kaplan-Meier curves.
What are Kaplan-Meier curves?
Kaplan-Meier curves were first introduced in 1958 to estimate the probability that a study sample would experience, or not experience, a particular outcome over time. In medicine, that usually means an adverse outcome, like a major cardiovascular event, cancer progression, or mortality.
Kaplan-Meier curves are presented as a graph, with the probability of not experiencing the outcome on the y-axis and length of time on the x-axis.
While Kaplan-Meier curves are most often used to measure mortality, their fundamental role is to measure time-to-event, which can be any event. In fact, Kaplan-Meier curves are used outside of medicine as well. A good example is a study from 2021 using the tool to study flight departure delays.
How to compare two Kaplan-Meier curves
In most clinical trials, two different Kaplan-Meier curves are displayed in the same figure as a comparison. They provide a visual representation of how much longer it takes for an event to occur in one group than in the other. One way of comparing the two curves, other than visually comparing them, is to use statistical tools to test if there are differences between them.
The log-rank test compares the survival over the entire curve between two groups. In doing so, it compares the observed values to what would be expected, given that the null hypothesis is true. In medical research, the null hypothesis is most often that there is no treatment effect.
It is also possible for Kaplan-Meier curves to cross over. This most commonly occurs when two interventions are being compared, rather than when a control is being compared to an intervention.
For example, if one treatment has a higher short-term mortality than another, but eventually leads to better overall survival, then the curves could cross each other. In this case, the log-rank test should not be used because the underlying assumptions of the test are not valid.
Communicating the information to your patients
Patients are likely most interested in Kaplan-Meier analysis after they are diagnosed with a terminal illness. They want to understand how long they might have left to live, based on their diagnosis.
One of the best ways to help patients understand time-to-event information is to ensure they understand how a treatment, whether it be curative or life-extending, would compare to no treatment. Doing so can make the information more approachable to patients.
Finally, it’s important for patients to understand that the results obtained were from groups of patients. So while we might know that treatment is better than no treatment, or that one treatment is better than another, we can’t say for sure how an individual patient will respond.
The bottom line
Why trust our experts?
Kaplan-Meier curves are an often-used statistical tool that helps determine the likelihood of experiencing an adverse outcome over a period of time. They also help us grasp the potential benefits or harms of treatment.
As a healthcare provider, it’s important to understand this test as well as how to communicate the information it provides to patients so they can make the best choice for themselves and their families.
