Author: Facundo Zaffaroni – Senior Biostatistician
Introduction
When running clinical trials, especially those for serious conditions, it’s not just about whether a treatment works but also when it works. This is where survival analysis comes in. Survival analysis helps researchers understand how long it takes for an event (like death, disease progression, or relapse) to happen in patients receiving a treatment.
In simpler terms, it looks at the time until something important happens and can compare it across different patient groups. This method is critical for understanding the effectiveness of new therapies, especially when the timing of outcomes matters as much as the outcome itself.
At IDDI, we work with pharmaceutical and biotech companies to apply survival analysis in a way that brings out clear, actionable insights for their clinical trials. In this blog, we’ll break down the basics of survival analysis in clinical trials and how it helps you make better decisions about your treatments.
What is Survival Analysis?
Survival analysis is a collection of statistical tools used to analyze the time it takes for an event to occur. In clinical trials, this event could be death, disease progression, relapse, or another critical outcome – it depends on the context. What makes survival analysis different from other types of data analysis is that it doesn’t just look at whether something happens… it looks at when it happens.
For example:
- Does the new treatment extend life expectancy?
- How long does it take for the disease to progress in patients receiving the new therapy?
This kind of analysis is especially important in trials for chronic or life-threatening diseases where time is a crucial factor in measuring the success of a treatment.
To help better understand the concept of ‘when it happens’: Imagine you and your friend order dinner (each from a different restaurant). You both plan to have dinner 45 minutes after ordering. Your friend’s food arrives precisely 45 minutes later, but yours took 2 long hours to arrive… While both dinners arrived, how impactful was that difference in time? Survival analysis specifically takes timing like that into account.
The Challenge of Incomplete Data
In most clinical trials, not every patient will experience the event by the end of the study. Some patients may drop out or the trial might end before the event happens. This is called censored data. We know that these patients haven’t experienced the event yet, but we don’t know when (or if) it will happen in the future.
Survival analysis takes these censored patients into account, so their data still contributes to the overall understanding of the treatment’s effects. This ensures that the trial results are as accurate as possible.
Key Methods in Survival Analysis
There are two main ways we look at survival data in clinical trials: Kaplan-Meier curves and the Cox proportional hazards model. Let’s break these down without getting too technical.
1. Kaplan-Meier Curves:
Kaplan-Meier curves are a “simple” and widely used tool to estimate the probability of survival over time. These curves give a visual picture of how long patients in different treatment groups survive without experiencing the event. For example, if you’re testing a new cancer drug, a Kaplan-Meier curve can show how the survival rates of patients on the new treatment drug compare to those receiving standard care.
The curves are easy to interpret – they can show how many patients have not yet experienced the event at different time points during the trial, while also providing the estimated probability of continuing without an event for longer periods. The further apart the curves are for different groups, the greater the difference in survival estimates between these groups.
2. Cox Proportional Hazards Model:
While Kaplan-Meier curves give you a visual idea of how different groups compare, the Cox model allows us to dig a little deeper. The Cox model helps answer questions like:
- How much does a treatment reduce the hazard of an event compared to the control, for example?
- Do factors like age or pre-existing conditions affect survival?
The Cox model generates something called a hazard ratio, which tells us how much more or less likely the event (like death, for example) is to happen in one group compared to another. A hazard ratio of less than 1 means the treatment is reducing the hazard of the event, while a hazard ratio of more than 1 means the hazard is increased when using the treatment.
For example, if a new heart failure treatment has a hazard ratio of 0.75, it means patients on the treatment have a 25% lower hazard of experiencing a heart attack compared to those on standard care.
IDDI Biostatistics Experts interpret survival analysis results in a way that aligns with your trial’s objectives, focusing on real-world implications of the data, making it easier to communicate results with your stakeholders.
Challenges in Survival Analysis: When Things Get Complicated
While the Kaplan-Meier curves and Cox models are incredibly useful, some clinical trials face more complex issues that require advanced methods. Here are a couple of situations where things get trickier:
1. Non-Proportional Hazards:
The Cox Proportional Hazards model assumes that the treatment effect remains constant over time. But what if this isn’t true? Sometimes, treatments might work better early on but lose effectiveness over time, or vice versa. When this happens, we talk about non-proportional hazards.
In these cases, standard survival analysis tools might not give the most accurate picture, so we use more specialized techniques to handle this.
2. Competing Risks:
In some trials, patients may experience more than one type of event. For example, in a study of a heart medication, patients could die from heart disease (the primary event of interest, for example) or from another cause, like cancer (a competing risk). Ignoring these other events can lead to misleading results.
Competing risks analysis helps us accurately measure the probability of the event happening when other risks are also in play.
At IDDI, we can help you spot these situations and apply the right methods to make sure your results are reliable.
How Survival Analysis Applies Across Different Therapies
While survival analysis is often associated with cancer trials, its use extends far beyond oncology. It plays a key role in trials for many other conditions where time-to-event is a critical factor. Here are just a few examples:
- Cardiology: Survival analysis helps determine how long patients remain free from major cardiovascular events like heart attacks or strokes.
- Infectious Diseases: It’s used to assess how long patients remain symptom-free or virus-free after treatment.
- Neurology: In diseases like Alzheimer’s or Parkinson’s, survival analysis can measure the time until cognitive or physical decline, helping to assess whether treatments slow disease progression.
Why Survival Analysis is Critical for Regulatory Approval
Regulatory bodies like the FDA or EMA often require survival data to evaluate the long-term effects of treatments. They want to see not only that a treatment works but also that it has a lasting impact on patient outcomes. Whether it’s extending life, slowing disease progression, or reducing relapse rates, survival analysis helps provide this critical evidence.
Conclusion: Making Survival Analysis Work for Your Trial
Survival analysis is a powerful tool that provides valuable insights into how and when treatments work. By understanding the timing of important events like disease progression or relapse, researchers and sponsors can make more informed decisions about the effectiveness of new therapies.
At IDDI, we specialize in making survival analysis straightforward and actionable. Whether you’re dealing with simple or complex survival data, our team is here to help you interpret the results and apply them in a way that improves patient outcomes and supports regulatory approval.
Need help with survival analysis in your clinical trial? Contact IDDI today to learn how we can assist you in getting the most out of your data!