How to Perform Overall Survival (OS) and Progression-Free Survival (PFS) Analyses in SPSS: A Comprehensive Guide

Estimated reading time: 5 minutes

• 2. Preparing Your Data for OS and PFS Analysis
• 3. Kaplan-Meier Survival Analysis
• 4. Running Kaplan-Meier Analysis in SPSS
• 5. Log-Rank Test for Comparing Survival Curves
• 6. Cox Proportional Hazards Model
• 7. Advanced Options in SPSS for OS and PFS
• 8. Visualizing Survival Data in SPSS
• 9. Checking the Assumptions of Cox Proportional Hazards
• 10. Practical Applications of OS and PFS Analysis
• 11. Common Pitfalls and How to Avoid Them
• 12. FAQs

Survival analysis is a powerful tool in medical research used to understand time-to-event data. Overall Survival (OS) refers to the duration from a defined start point, such as diagnosis or treatment, until death from any cause. In contrast, Progression-Free Survival (PFS) is the time from the same start point until disease progression or death. These metrics are crucial in clinical trials, especially in cancer research, where treatments aim to prolong survival or delay disease progression.

In survival analysis, it is essential to account for “censored” data — cases where patients haven’t experienced the event of interest (e.g., death or progression) by the end of the study period. SPSS is a widely used statistical software that allows researchers to perform survival analysis through functions like Kaplan-Meier and Cox regression.

2. Preparing Your Data for OS and PFS Analysis

Before performing survival analysis in SPSS, ensure your dataset includes two crucial variables:

• Time Variable: This represents the duration (in days, months, or years) until the event occurs.
• Event Status: This binary variable indicates whether the event has occurred (e.g., 1 for death or progression) or if the data is censored (e.g., 0 for no event).

Make sure your dataset is structured appropriately for SPSS. Each row should represent an individual subject, with separate columns for time, event status, and any covariates (e.g., treatment type or age).

3. Kaplan-Meier Survival Analysis

The Kaplan-Meier method is the most common way to estimate survival probabilities. This non-parametric method calculates the probability of surviving past certain time points, providing a step-function graph known as the survival curve.

Steps to Perform Kaplan-Meier Analysis in SPSS:

1. Go to Analyze > Survival > Kaplan-Meier.
2. Set the time variable and event status.
3. If you want to compare survival between different groups (e.g., treatment types), define a grouping variable.
4. Click OK to generate the Kaplan-Meier curve.

The output will include the median survival time and confidence intervals for each group, allowing you to visually compare survival across groups.

4. Running Kaplan-Meier Analysis in SPSS

Once you’ve prepared your data, SPSS will produce survival curves that visually represent survival probabilities over time. These curves are typically stepped due to the discrete nature of survival events occurring at specific times.

5. Log-Rank Test for Comparing Survival Curves

The Log-Rank Test is used to determine whether there are statistically significant differences between survival curves of different groups. This test compares the entire survival experience of groups over time.

In SPSS, this can be done by checking the option to include the Log-Rank Test when performing Kaplan-Meier analysis.

6. Cox Proportional Hazards Model

While Kaplan-Meier provides an estimate of survival probabilities, the Cox Proportional Hazards Model allows you to examine the relationship between survival time and multiple covariates (e.g., age, treatment, tumor size). The model estimates the hazard ratio, which tells you the risk of the event occurring at any point in time.

Steps to Perform Cox Regression:

1. Go to Analyze > Survival > Cox Regression.
2. Set your time and event variables.
3. Add covariates such as age, gender, or treatment type to the model.
4. Click OK to run the analysis.

7. Advanced Options in SPSS for OS and PFS

Handling censored data correctly is crucial in survival analysis. SPSS allows you to account for censoring by indicating which subjects have not experienced the event by the end of the study. The software also offers options to perform stratified analysis, where you can analyze survival for different subgroups (e.g., male vs. female).

8. Visualizing Survival Data in SPSS

SPSS generates Kaplan-Meier survival plots, which can be customized to display different groupings and highlight key time points. These plots are an essential part of survival analysis, offering an easy way to communicate the results visually.

9. Checking the Assumptions of Cox Proportional Hazards

The Cox model assumes that the hazard ratio remains constant over time. It’s essential to test this assumption before drawing conclusions. SPSS includes diagnostic tools to check whether this assumption holds true.

10. Practical Applications of OS and PFS Analysis

Survival analysis, particularly OS and PFS, is vital in clinical trials to assess the efficacy of treatments. In oncology, for instance, researchers use OS and PFS to determine how well a new drug delays tumor progression or extends patient survival.

11. Common Pitfalls and How to Avoid Them

• Misinterpreting Survival Curves: Ensure you understand the difference between survival probabilities and cumulative risk.
• Incorrect Use of the Log-Rank Test: The test should only be used when comparing two or more survival curves.
• Data Censoring Errors: Make sure that censored cases are appropriately marked to avoid skewed results.

12. FAQs

Q1: What is the difference between OS and PFS?
A1: OS measures time to death from any cause, while PFS measures time to disease progression or death.

Q2: Can I include multiple covariates in Cox Regression?
A2: Yes, Cox regression allows for multiple covariates to assess their impact on survival.

Q3: How do I handle censored data?
A3: In SPSS, censored data should be coded as 0 in the event status variable.