AstraZeneca
Contact: Yoshiharu Horie, PhD
Country: Japan
Award Category: Innovative Problem Solver
Tell us about the business problem you were trying to solve.
AstraZeneca (AZ) is committed to create innovative medicines that contribute to the patient health. As multinational randomized clinical trials have become the standard for new drug development in recent years, the participation of Japanese patients has been limited. Therefore, there is not sufficient information on the efficacy and safety of new drugs for the Japanese patients before launch. In the meanwhile, Real World Data (RWD) from routine clinical practice of patients contains extensive information on diagnosis, treatment and procedure. Its use has been significantly increasing year by year in the healthcare industries. AZ actively promotes the use of RWD from the clinical development stage through post-marketing.
AZ conducts database (DB) studies to deliver valid evidence to patients and healthcare professionals. This approach ensures we obtain highly accurate RWD analysis results and distribute them rapidly. It leads to clarify the status of the current therapy, guideline revisions, and raising disease awareness.
The volume of RWD is huge, typically greater than 50 GB. In one project as an example, we analyzed the RWD which including patients with Gastroesophageal reflux disease (GERD), the dataset consists of approximately 3.5 million patients and 5 billion records, making around 200 GB as SAS datasets.
As our mission is to publish research with high-quality scientific evidence, it is critical to store valuable patient data with high security and analyze it using appropriate methods within a strict timeline. Also, the cost of person time spent in the project is controlled strategically. If the costs become too high, the number of DB studies that can be conducted will be limited, making it difficult to contribute to patient health care.
What SAS products did you use and how did you use them?
To address this challenge mentioned above, we have decided to tackle the issues faced by our RWD Data Science members by using SAS Viya on our cloud environment Japan Data Hub (JDH). As we have been using SAS, the standard in the pharmaceutical industry, for many years, we have decided to utilize SAS Viya with the new processing engine, CAS (Cloud Analytic Services). Specifically, we explored the possibility of reducing the processing time for large datasets by using CAS instead of relying solely on the conventional SAS9 compute engine, SPRE (SAS Program Runtime Environment).
The process was improved from the following perspectives:
1: Standardization of Analysis Environment and Process
Primarily, the analysis time can be managed consistently across multiple DB studies, and it also enables timely verification of the dataset and analysis specifications and programs, as well as quality control for the reuse of the program.
2: Reutilization of analysis program and analysis result formats
Data scientists, statisticians, and members of Clinical Research Organizations (CRO) can analyze RWD at the JDH. Then, the formats of all analysis results were standardized regardless of the type of studies, and analysis programs were centrally managed there, allowing for reuse to enhance efficiency and quality.
What were the results or outcomes?
We achieved positive results in terms of productivity and cost efficiency via this initiative to tackle the challenges we mentioned earlier.
A. Reduce Analysis Time
Before using SAS Viya at the JDH for analysis, the analysis period (from finalization of the study protocol to output of the analysis results) was approximately 10 months. As the results of using SAS Viya and standardizing the process, we successfully reduce the analysis period to about 6 months.
B. Cost Reduction
Previously, the cost per DB study exceeds 150,000 USD, we achieved a cost reduction of approximately 60% (a decrease of 90,000 USD) by analysis process standardization mainly using SAS Viya at the JDH.
One of the key factors contributing to the outcome is the substantial reduction in processing time. By performing data processing (e.g., sorting and deduplication) using CAS, the data processing time was reduced significantly from 4 hrs 37 mins to 2 hrs 29 mins. Additionally, the time of reading RWD from SAS datasets was shortened from 4 hrs 19 mins 24 secs to 2 hrs 34 mins 38 secs. (for the dataset of 203 GB with 520 million records).
Notably, it is often expected to require multiple iterations of data wrangling tasks through trial and error when preparing an analysis dataset in the DB study. As assumption, this data preparation process is repeated about 50 times in a single DB study, it was estimated to save the time equivalent to 35 business days.
Why is this approach innovative?
SAS 9 has been used for data analysis (mainly new drug application for clinical trials) in the pharmaceutical industry for a long time. However, in the past, it was impossible to analyzing RWD with past technology, which often involves huge amount of data.
By fully leveraging the power of latest SAS technology, we have made it possible not only to conduct accurate analysis of a huge amount of data such as RWD but also to reuse analysis programs in a sustainable manner.
In recent years, AZ has been focusing on three key areas to achieve sustainability: "Access to Healthcare", "Environmental Protection", and "Ethics and Transparency." Our initiatives are innovative because we enable the sustainable reuse of analysis programs to generate evidence faster to facilitate patient’s access to healthcare.
In conclusion, we aim to efficiently utilize RWD of Japanese patients and enhance evidence generation process to realize patient-centered evidence creation while maintaining our commitment to sustainable activities through our innovative approach.
