In this podcast, Dr Juliana Tolles discusses time-to-event analysis in clinical research. She explores the concept of survival curves, Kaplan-Meier curves, and statistical tests to compare survival rates. She also explains the differences between the log-rank test and the Cox proportional hazards model. Additionally, she analyzes the proportional hazard assumption and provides an example study comparing major adverse cardiac events.
Time-to-event analysis compares rates of positive or adverse events, even if events are not observed during follow-up.
Survival curves, Kaplan-Meier plots, and statistical tests like log-rank test and Cox proportional hazards model are used to compare groups and quantify risk factors in survival analysis.
Deep dives
Time to Event Analysis: An Introduction
In this podcast episode, the hosts discuss time to event analysis, also known as survival analysis. The analysis compares the rate at which patients experience different outcomes, such as positive or adverse events. One challenge in analyzing this type of data is that patients may be followed for different periods of time or may be lost to follow-up. Survival data can arise in various forms in medicine, reflecting the time until a patient experiences an event, whether positive or adverse. The key point is that even if an event is not observed during the follow-up period, the fact that the event didn't occur provides valuable information.
Understanding Survival Curves and Kaplan-Meier Plots
The hosts explain how survival curves and Kaplan-Meier plots are used to represent the fraction of patients who do not experience an event over time in different treatment groups. These plots visualize the likelihood of surviving to each time point and are a common feature in clinical research reports. Additionally, statistical tests like the log-rank test and the Cox proportional hazards model are used to compare survival curves between groups. The Cox proportional hazards model allows researchers to account for independent variables that might impact the outcome, while the log-rank test focuses solely on comparing survival.
Interpreting Hazard Ratios in Time to Event Analysis
The hosts discuss the concept of hazard ratios, which quantify the magnitude of risk associated with a particular factor. In the context of survival analysis, hazard ratios represent how a specific treatment or risk factor influences the chance of experiencing an event compared to the baseline risk. The example study by Nissen et al. comparing weight loss treatment to placebo for major adverse cardiovascular events is examined. The study concluded that the hazard ratio associated with the active treatment was less than 2, indicating that the treatment did not significantly increase the risk of adverse events. However, due to the non-inferiority study design, conclusions about a hazard ratio less than 1 could not be made.