Efficient Change-from-Baseline Calculation in Clinical Trials Using SAS® Hash Objects

Photo by Joshua Aragon on Unsplash

Introduction

Clinical trials play a crucial role in the development and evaluation of new therapeutic interventions for various health conditions. The primary focus of many clinical trials is to assess whether the treatment groups differ in terms of the change from baseline to the end of therapy in a continuous response variable. To achieve this goal, researchers employ a repeated measures design, where subjects are measured at fixed time points throughout the study. This article will discuss the process of calculating change from baseline in SAS® and introduce the hash object as an alternative method for efficient change-from-baseline calculations in clinical trials.

Traditional Method for Computing Change from Baseline

Traditionally, the most common method for computing change from baseline in SAS® involved dividing the original dataset into two subsets: baseline and post-baseline data sets. The baseline value is typically derived from a single point in time, often the last value measured before the first dose of the study drug, or it may be a composite of several observations, such as the mean of the last two measurements before the start of the study drug administration.

Once the baseline value is determined for every subject, a separate dataset containing this information is created. This dataset is then merged with the post-baseline dataset by subject and diagnostic parameter. Finally, the change from baseline is calculated for each patient and parameter. While this method has been widely used in the past, it can be time-consuming and computationally inefficient, especially for large datasets.

Introducing the Hash Object in SAS® 9.2

SAS® 9.2 introduced the hash object as a new method for quickly calculating change from baseline values. The hash object is a powerful data management tool in SAS®, allowing users to store and retrieve data in memory without the need for sorting or merging datasets. This results in significant improvements in processing time and computational efficiency compared to traditional methods.

Hash Object Syntax and Basic Concepts

To work with hash objects in SAS®, users need to understand the basic syntax and concepts. A hash object is created using the ‘declare’ statement, followed by the name of the object, and ‘hash’ keyword:

declare hash object_name();

The dataset to be loaded into the hash object is defined using the ‘defineKey’ and ‘defineData’ methods. The ‘defineKey’ method specifies the key variable(s) to be used for indexing the hash object, while the ‘defineData’ method specifies the data variable(s) to be stored:

object_name.defineKey('key_variable(s)');
object_name.defineData('data_variable(s)');

The ‘load’ method is used to read the dataset into the hash object:

object_name.load(dataset: 'input_dataset');

Once the data is loaded into the hash object, it can be accessed and manipulated using various methods, such as ‘find’, ‘add’, ‘replace’, and ‘remove’.

Calculating Change from Baseline Using Hash Objects

To illustrate the use of hash objects for calculating change from baseline in a clinical trial, consider a dataset containing patient IDs, diagnostic parameters, visit dates, and measured values. The goal is to calculate the change from baseline for each post-baseline visit for each patient and diagnostic parameter.

First, create a dataset containing only baseline measurements:

data baseline;
    set input_dataset;
    if visit_date = 'baseline';
run;

Next, create a hash object and load the baseline dataset into it:

data change_from_baseline;
    set input_dataset;
    if visit_date ne 'baseline';
    /* Declare and load the hash object */
    declare hash baseline_hash();
    baseline_hash.defineKey('patient_id', 'diagnostic_parameter');
    baseline_hash.defineData
('baseline_value'); baseline_hash.load(dataset: 'baseline');
/* Retrieve baseline value and calculate change from baseline */
rc = baseline_hash.find();
if rc = 0 then do;
    change_from_baseline = measured_value - baseline_value;
end;
/* Output the change from baseline data */
output;
/* Clear the hash object */
baseline_hash.clear();
run;

In this example, the baseline dataset is loaded into a hash object named ‘baseline_hash’, with ‘patient_id’ and ‘diagnostic_parameter’ as key variables and ‘baseline_value’ as the data variable.

The ‘find’ method is then used to retrieve the baseline value for each post-baseline record in the input dataset. The change from baseline is calculated by subtracting the baseline value from the measured value at each post-baseline visit.

Advantages of Using Hash Objects

The use of hash objects for calculating change from baseline offers several advantages compared to traditional methods:

1. Speed and efficiency: Hash objects allow for fast in-memory data processing, which can result in significant improvements in processing time and computational efficiency, especially for large datasets.
2. Flexibility: Hash objects can be easily modified and manipulated, making them an ideal tool for handling complex data management tasks in clinical trials.
3. Reduced need for sorting and merging: Hash objects eliminate the need to sort and merge datasets, which can simplify the data processing workflow and reduce the risk of errors.

Conclusion

In clinical trials, calculating change from baseline is an essential step in evaluating the effectiveness of a therapeutic intervention. The introduction of the hash object in SAS® 9.2 has provided an efficient and flexible method for quickly calculating change from baseline values. By leveraging the power of hash objects, researchers can streamline their data processing workflows, reduce computational overhead, and focus on drawing meaningful conclusions from their clinical trial data.