XLSTAT: Kaplan-Meier Survival Analysis in Excel

 

Dataset for Example

The attached data, from Gehan (1965), come from a clinical trial assessing the effect of 6-mercaptopurine on leukemia remission times.

The study compares survival curves of two groups: 21 treated patients and 21 controls. All control patients had recurrences, while in the treated group, 9 had observed recurrences and 12 were censored. The goal is to evaluate the drug’s impact on survival time.

 

Setting Up

• Open XLSTAT.

• Click on the Life Sciences/ Survival analysis menu and select the Kaplan-Meier analysis feature.

• After clicking the button, the Kaplan-Meier analysis box opens.

  In the General tab:

  • Select the data from Excel.

  • Set Time data as the duration until relapse or censoring.

  • Set Status indicator to 1 for relapse (event) and 0 for censored.

In the Data options tab:

• Activate By group analysis and select the group variable to distinguish control and treated patients.

• Enable Compare to perform group comparison tests.

• Use Filter to select specific groups for analysis if needed.

In the Charts tab, activate desired charts and select a symbol to represent censored data.

• Click on the OK button to launch the computations.

• The results are displayed on a new Excel sheet.

 

Interpreting the Results

The first table displays a summary of the data for each group. Then XLSTAT displays results specific to each group. It starts with the Control group.

First, the Kaplan-Meier table is displayed. It contains the results of the Kaplan-Meier analysis with several key indicators such as the Survival distribution function.

The next tables give the mean and median survival time and the respective confidence intervals.

Then, we can visualize several curves, including the survival distribution function (SDF, or survivor function, or reliability function), bounded by the confidence intervals.

Next, the same series of results is displayed for the Drug 6-MP group.

The following results contain some values that are missing because they could not be computed.

We notice that the median survival time is a lot lower for the control group than for the 6-MP group (8.667 vs 21.943).

In the SDF, circles identify the censored data.

Then, we can compare the two groups. First, a series of tests is displayed in a table (Log-rank, Wilcoxon, Tarone-Ware). From the results we can see that the difference between the two survivor functions is very significant.

Last, the comparison of the two survival curves allows us to conclude to confirm that the drug significantly improves survival time of patients.