Creating a Kaplan-Meier Analysis Table Using ADaM Datasets
Clinical trials are a critical part of the drug development process, providing important data on the safety and efficacy of new treatments. As part of the clinical trial reporting process, statistical analysis plans (SAPs) are developed to outline the statistical methods that will be used to analyze the data. One common type of analysis is a Kaplan-Meier analysis, which is used to estimate the probability of survival or time until a specific event occurs.
In this article, we will focus on the creation of a Kaplan-Meier analysis table using ADaM datasets. Specifically, we will use the example of the efficacy parameter described in Section 4 of the SAP as “Time until the first diastolic blood pressure ≤ 90 mmHg is achieved.” We will walk through the process of creating the necessary datasets and variables, as well as discussing some of the challenges that may arise during the analysis.
To start, we need to review the SAP to understand the requirements for the analysis. In Section 6.4, we find the definition for the time until the first diastolic blood pressure is reached. We also learn that we need to censor subjects who do not achieve that by Week 24 and that we need to censor subjects who discontinue prior to Week 24 at the last date read. Additionally, we will need to create a variable called CNSR (censor) to identify how many subjects were censored for each group.
Next, we will need to create the necessary datasets to support the analysis. The Analysis Data Sets for Vital Signs (ADVS) dataset provides a structure for capturing data such as blood pressure readings. However, we will need to create additional variables to support the Kaplan-Meier analysis. Specifically, we will need to create a variable to identify the date on which each subject achieves the ≤ 90 mmHg reading, as well as a variable to identify the number of weeks between when a subject first took the medication and when their diastolic blood pressure reached 90 mmHg.
One of the challenges with Kaplan-Meier analyses is the need to account for censoring. Subjects who fail to achieve a 90 mmHg diastolic pressure or who do not follow the protocol requirements may still be counted, but their results will be censored. We will need to identify these subjects in our analysis and account for them in the final table. This may involve creating a table or listing showing all of the reasons subjects were censored.
Another challenge with Kaplan-Meier analyses is the need to track which visits are being used to determine the number of weeks until the event. For example, if the SAP specifies that only the first date is to be used where two subsequent diastolic readings were at or below 90 mmHg, we will need to track which record to use. This is important not only for the statisticians but also for regulatory agency reviewers and for debugging purposes.
Once the necessary datasets and variables have been created, we can move on to performing the actual analysis using statistical software such as SAS or R. This involves applying appropriate statistical methods such as the Kaplan-Meier estimator to estimate the probability of survival or time until the event occurs. The resulting table will include key variables such as the number of subjects at risk, the number of events, and the probability of survival or time until the event occurs for each treatment group.
In conclusion, Kaplan-Meier analysis tables are an important part of clinical trial reporting, providing critical information on the probability of survival or time until a specific event occurs.
